Re: Why year 1904 is chosen for start date?
From: Ben Combee [EMAIL PROTECTED] At 09:53 AM 10/18/2005, you wrote: As for why that date exactly I think it's related to leap year calculations. By avoiding 1900 they avoided some possibly complicated leap year issues (only to run into them again in Y2K, of course). Actually, the leap year problems did not occur with 2000, since it was a multiple of 400 and had a leap day -- 1900 didn't have a leap day, so handling it would have been a problem. Great in theory, except for those programmers that did enough research to find the 'divide by 100' rule and not enough to find the 'divide by 400' rule. D'oh Not that this has anything to do with Palm OS :-) Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: odd address error
From: cbruner [EMAIL PROTECTED] (as part of a possibly off-topic discussion about structure byte-alignment) ... seems to me to be very byte centric. The only difference between reading at an even boundary and odd one is one bit. The as far as the boundaries being set up as dwords/words/bytes is just a matter of convenience, mainly for programmers. Nope, it's _in_convenient for programmers, done purely for processor speed. Even the processors that allow non-word oriented data explicitly document this as slow and to be avoided. And, since it generally can be avoided, anyone that packs a structure array to an odd offset is producing a bad design because this has an unknown speed and code complexity hit. Hell, half of the quad-byte aligned structures out there have a chance of failing on the new 64-bit processors, a google search will show you the problems people are having with alignment so it's not something you want to avoid thinking about. After all, if you're accessing an indexed store written by a different processor issues such as endianness and byte-alignment have a fair chance of making your goal of mapping the processor-independent disk-based index directly to an in-memory array optimistic and unlikely. I agree that sometimes you get handed stuff designed by someone who didn't think past the end of their desk but I'd be more likely to point the 'bad design' finger at whoever handed you that data structure, not the designer of the processor you happen to be compiling for at the moment. Me, I'd have built the index as two arrays, one with the keys and one with the offsets, each of which would then be correctly aligned for whichever processor it was loaded onto. Although I guess that is just as likely to introduce some sort of problem somewhere, but that's optimisation for you. It's never free. Are we off-topic for this list yet? :-) Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Compressing records with Lz77
From: Trevor MacPhail [EMAIL PROTECTED] I'm writing an application that is going to be working with a potentialy large database of almost all text and I'd like to compress the records. Is it possible to use PalmOS's Lz77 API for this? And if so, HOW? I can't find any documentation of the functions other than what little there is in the Lz77Mgr.h header file. I successfully used the zlib stuff from zboxz http://palmzlib.sourceforge.net/ to add compression to a project. My project was cross-platform and I had to rework the libraries to embed them but even so I'm happier I started there than starting from scratch. No idea how this compares with the Lz77 API. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: command line building for metrowerks
From: David Birdsall [EMAIL PROTECTED] According to the manual for codewarrior (...) you can script codewarrior to do automated builds and the like. It talks about using the OLE/COM object viewer to look for interface definitions of objects you can script using vbs/jscript (using wscript.exe on windows at least). There are some sample scripts included with the CW release (folder name of (CodeWarrior Examples)\Scripting Examples) that were easy enough to follow. They are documented and include code to copy the compiler messages to a log file. I found it fairly easy to extend the basic build script to read a file of projects in order to automate the building of a set of libraries and the apps that used them, although if I had to do it again I'd start by finding a vbscript debugger. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Nested_FrmPopupForm().
From: rci_ind [EMAIL PROTECTED] Can i use nested FrmPopupForm(). (snip of 'form A, form B, form C sequence) I don't think so but, more importantly, a form sequence this complex is going to be difficult to debug and maintain. It's not generally the way complex Palm interfaces do things. How should i ensure the proper closure of formB and formC? will i still have Field1 undisturbed? If you need to spread your data management over several forms you are better to isolate your data saving and loading to a separate module and then have each form interact with that. Each form then uses those routines to load the current record (generally using an index stored in a global variable), interacts with the user to edit the data and then saves the fields it has modified. In theory this means that the data module is free to save all of the data back to the database each time one of the forms closes, or to keep it all in global memeory until a final 'Save' button is hit and then update the database record. Your forms don't have to manage the data themselves and so it becomes easier to do things such as move a field from one form to another or split or combine forms later on. The data module can be built as if it was an object with get() and set() methods for each field. Although C doesn't support object-oriented programming syntactically there's no reason you can't apply object-oriented design principals. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: The File vs Database Paradigm
From: Lance Drake [EMAIL PROTECTED] My application creates records. I do not want to create PDB files because the destination for the records is FileMaker or Excel - which don't know how to read those Palm binary file formats for import into records or spreadsheets. Yes, I could write a plug-in for the destination apps - but that seems riskier (because of assumptions about the PDB internal structures) than just making the XML text available. One option is to tag the databases for backup and then look for them in the desktop backup folder after the hotsync has completed and use a pdb2csv converter to produce files that desktop apps can import. It's an alternative, although the last time this was discussed there were a few people pointing out that conduits have become _much_ easier to write (conduits also don't require knowing anything about the PDB internal structures - the conduit API does that for you). Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to find a bug that crashes Simulator?
From: Aaron Hansen [EMAIL PROTECTED] Is this not an issue for anyone else? I'd at least like to know if other people are able to trap fatal errors in any simulator without it GPF'ing too. It's not like the old days when I had the source code for the POSE and I could just throw it in the debugger and track down the crash myself. Or is it? Is the source for any of the simulators available? My experience has been that a crash in the simulator was usually caused by a bug in my code resulting in either memory corruption or an API call being passed an invalid value. The simulator, because it's just a copy of the normal Palm OS ROM, can't handle this and dumps out. This, unfortunately, makes it a much less powerful debugging aid than the emulator (which could tell you about stuff like null pointers and memory corruption) but it's a lot cheaper to build this way for PalmSource. The bottom line is that you first need to gaurantee that your code isn't causing the crash and to do that you've got to employ the same sort of strategies as debugging on the device - log to the screen, log to a db, log to a VFS file, etc. The crash is far more likely to be a bug in your code than a bug in the Palm OS source and, if it is a bug in the Palm OS source you're down to debugging assembler because you're not going to get access to the Palm OS source. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to find a bug that crashes Simulator?
From: Aaron Hansen [EMAIL PROTECTED] I never suggested I want to find bugs in the PalmOS code. I want to find bugs in my code. Of course we try to garuntee that we never generate a memory exception or use an API incorrectly, but that is going to happen with a code base as large as the one I work with (60 segments to give you an idea). There are several developers touching the same code, and some of it is code that is common to other plaforms like Symbian, PPC and SmartPhone. When it does happen the simulator should not GPF. No amount of discussion will convince me that is OK. You've got my full agreement there. I was amazed at the simplicity of some of the crashes in terms of bad API parameters. How hard would it be for Palm Source to provide a developer build that left the ASSERTS in place and generated a stop of some sort. At least then you could quickly identify the location of the crash. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to find a bug that crashes Simulator?
From: Ben Combee [EMAIL PROTECTED] The simulator is a whole build of the OS with a Windows-based device abstraction layer. It's used for OS development at PalmSource, so usually a crash in the sim is detected in the debugger of Visual Studio by a developer that has full simulator source code. No source is available, because the sim source is actually the whole OS source. Excuse me? It's used for OS development at PalmSource. My mistake then, because I thought it had been released as a tool for Palm OS developers. It's one thing to replace the robust, well built and useful emulator with a simple simulator - cheap, but that's life. It's something else to respond to complaints by suggesting that it's only an internal tool. If it's an internal tool then Palm Source need to identify it as such and, on that basis, admit that OS 5 has no debugging environment. If it's not an internal tool but is intended for use as a debugging aid for third-party developers then let's have a debug build that doesn't crash when an ASSERT fails (I'm assuming that the crash is an ASSERT and not a simple failure to check API parameters). Which puzzles me because, as I'm typing this, I'm remembering that there *is* a debug version of the simulator and debug ROMS which, if I remember correctly, crashes just as badly as the 'release' stuff. What's the difference between the debug and release versions? I've also just noticed an early release of an ARM emulator for Palm OS 5.4 and 6. Is that intended to replace the simulator as the primary debugging tool or is it just for low level tech stuff? Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to find a bug that crashes Simulator?
From: Ben Combee [EMAIL PROTECTED] Make sure you're using a debug version of the simulator. That has a lot more checks, since the assertions are turned on in the OS code. When one it hit, you should be able to hit the debug in 68K button and see your stack crawl. If someone identifies a crash as relating to a situation where an assert seems to be missing (perhaps a bad API parameter crashing the simulator rather than triggering an assertion) is there any value in raising this as a bug, or would it be ignored because it doesn't affect the real-world behaviour of Palm OS? I realise that it's a gray area, but I'm thinking that this at least would allow us to gradually improve the robustness of the simulator as a debugging tool, although it depends on PalmSource being willing to apply resources to those bug reports. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Fatal Exception on Zire 71 and 72
From: [EMAIL PROTECTED] i´m working on this application that was tested originally on Tungsten C and Sony Clie. In both was working fine. But now our customer will install like 100 Zires 72 with this application. The problem is that when trying to exit the application there is a fatal exception. I've tried to debug with CW, but I can not reproduce the fault. The same Zire works perfect while debugging. I have no idea what the problem is, but found that an excellent tool for tracking down this sort of problem was a simple logging module that wrote text to a PDB. Although this isn't as good as source-level debugging it has the advantage of running on the actual device and so has more chance of capturing the sequence of events that lead to the crash. I had a fairly complex module that tracked many resource issues for tracking problems with serial-connected hardware, but for this sort of thing all you really need is a couple of functions that - create a database using a timestamp for the name (so that new runs don't over-write previous debug data and to help keep them organised when you've hotsynced them back to your PC). - a function to write a string to the pdb (it can be useful to format the string as a string of comma separated values that starts with an incrementing index and then a tickcount - this lets you easily import it into a spreadsheet to filter out the crap). - then call that function at the top of important functions and anywhere you want to dump a data value or signal that something happened. Hotsync the device, grab the pdb from the backup directory and use one of the pdb 2 cvs converters to convert to a text file (one of the reasons you want each record to just be a formatted string) and voila - you can see the point the code crashed. It'll take a couple of round trips to start isolating the problem but with any luck it will get you to the point where you can actually get some info about the crash where-as at the moment you're probably just trying to guess what might be happening. It sounds like a lot of work just to trace the execution flow but trust me that this is invaluable (and reusable) whenever you've got a problem that only occurs on the device (or that other support nightmare: a problem that only occurs on one client's device). One of these days I'll get around to putting my debug module up on-line, but I just don't have the time to debug and document it. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: problem adding rsrc to mcp project
From: Ben Combee [EMAIL PROTECTED] The palms applications here were created in PILRC and some in constructor. I am trying to bring it uniform to use one resource editing method (don't tell me constructor is crap, it's what I'm usingJ). CW V9 uses PilRC as the standard format for projects using the Palm OS 68K Linker. To use Constructor resources, you need to create projects using the stationery, not the wizard, as that uses the Mac OS 68K Linker and PalmRez which allows Constructor-format resources. It's also not too hard to support both styles in the same project (which can be useful for testing during the transition). You have to take a project built for one style and manually set up the file types for the other. Sorry to be vague, but it's several years since I did this. You also end up with two different sets of resource setting locations (the new dialog inside the CW project IDE and also the constructor settings) so you have to be careful what you mix and match. I had a project with constructor-built forms and app definition but string resources and bitmap families created in the PilRC format via a custom-built app (for managing internationalisation with run-time language selection). It takes a bit of setting up and generally isn't necessary or desirable so I'm only mentioning that it's possible as an interim measure. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Duplicating an asp database web tool
From: Ornstein, Adam [EMAIL PROTECTED] 1. (brute force?) I was thinking that we could spider the tool, and produce a large number of files, that could be linked to one another from the PDA tool's main interactive page. Making it seem interactive... but is there something out there that can make that happen? Regarding html apps on the Palm, I've had some conversations with these guys and they might have an option for you. Copy of an earlier forum message From: James Fisher [EMAIL PROTECTED] Our viewer included on the ROM of the T5 and LifeDrive, used to render Addit has a lot of functionality that makes it useful as an html front end for 3rd party applications. We have been planning to publicly release our SDK to allow developers to take advantage of the viewer within their applications but the lack of business case means it keeps getting pushed back. (snip) If you are interested in playing with the SDK within your application contact me offline. The Addit conduits will also be available to developers for use within their application for system and other updates. James Fisher Bluefish Wireless -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: OS 5 is messing up certain numbers
From: Jim McGowen [EMAIL PROTECTED] Found something strange... Looks like Palm OS 5 changes certain numbers multiplying. For example: double x = 0.141; long y = (long)( x * 1.0 ); y becomes 1409. Also: double x = 1410.0; double test = 0.141; double y = test * 1.0; double z = 0.141 * 1.0; x = = 1410.0 = = 0100100101101000 y = = 1410.0 = = 0100100101100111 z = =1410.0 = = 0100100101101000 This is just a rounding issue isn't it? 0.141 * 1.0 has a good chance of being resolved at compilation time rather than run time so it's just an assignment of a precalculated constant where the test * 1.0 is subject to the accuracy with which that number can be stored and computed. To my knowledge casting isn't a good way to obtain an accurate conversion of a floating value because you don't have control over issues like truncation. I think you should be using an explicit conversion function so that 1409.99... is rounded to 1410 rather than being truncated to 1409. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Need some help determing what might be crashing this code
From: Henk Jonas [EMAIL PROTECTED] druid wrote: Hmmm and what determines poor coding practices. I guess, Roger was rather talking about your C-code style and all the bugs you put in your code and then ask here for help. Dude, you should better watch your code carefully, use a debugger and debug output and try to remove all the obvious mistakes before asking for help every day in the forum. Well the way I read it he was talking about the use of a global pointer to store the pointer of an open database. I do something similar, although wrapped in a module so that each database has a module that manages it, but some people are agressively opposed to global variables and wouldn't like my approach at all. But it was a fairly vague statement so... As to the 'obvious mistakes' I'm not sure how many mistakes are obvious when you don't know the answer and druid at least shows signs that he's investigating these problems himself. We've had posters in the past that have said I can't do this, please write the code for me and I need it by tomorrow. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Database: memory reading error
From: Stephan [EMAIL PROTECTED] Hi all! I am implementing a database with a list and edit tables similar to the address example. My problem is after entering text in the second field in my dataColumn I receive an error (also attached) that my program is reading from a location in the memory manager’s data structures. Any help will be appreciated!! And an excellent error message it is. The last sentence about writing past the end of the buffer says it all. Check the logic that limits the amount of data written and the size of the buffers allocated. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Need some help determing what might be crashing this code
From: druid [EMAIL PROTECTED]
Run it twice in a row and it will crash
EditRecord()
{
(snip)
err = Open();
h = DmGetRecord(dbPtr, nRecordIndex);
(snip)
MemHandleUnlock(h);
DmReleaseRecord(dbPtr, nRecordIndex, true);
err = DmCloseDatabase(dbPtr);
return 0;
}
You're finishing with a DmCloseDatabase but it's not
clear from this code whether Open() opens the database
or returns a stored pointer. Good practice is to allocate
and free resources at the same level so if you've got an
Open() function that does some special database management
you should also have a Close() function to reverse this. Of
course if Open() does something else entirely then your bug
is elsewhere.
Chris Tutty
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: How to I load a text file into a Database?
From: John Spence [EMAIL PROTECTED] The problem is that the database itself needs to be updated externally about 4 times a year. The update form is simply a new set of data in a Text file in Comma Separated Value form. The update file is more than 64K in length. There's no problem with the database size (details below) but the newer Palm OS storage systems have brought back the old problem with small record sizes so it's more important to make sure that your records are a reasonable size - several hundred bytes is fine. For more info read up on database bloat problems in the forum archives. a) Upload the text data file to the Treo650? Does the file need to be converted before the Upload? b) Load the data in the text file into a standard Palm Database? c) Handle the issue of the large amount of data (greater than 64K)? The best approach, IMHO, is to convert this to a Palm database (PDB) file on the desktop so that you can just hotsync it to replace the existing database. There are a number of products that will do csv to pdb conversion. Google or an forum archive search should point you in the right direction. Because you're converting to a PDB on the desktop you're not trying to fit the whole data block into a single memory object so 64K becomes completely irrelevant (as long as you're not talking about a database with only one record). Multi-megabyte databases are common with Palm apps. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: preload a field on a popup form
From: druid [EMAIL PROTECTED] Can some one explain how I preload a field on a popup form so when I call that popup form from the main form the field is already loaded with the records field data The field doesn't exist until the form is created. Note that although most forms are created and displayed in the same call that you can create a form in memory, use that form in API calls and then display it later (the FrmInitForm() call in Robert's code). This gives you a couple of options when it comes to situations where you have data available that isn't currently displayed: 1. Create the form when the other data is read but don't display it. This gives you the chance to prepopulate the form and keep it stashed away ready for display, but seems a bit clumsy to me. This is also more difficult because you've got to manage the form creation and disposal yourself and there are a number of different problems that can arise however if you're using popup forms you've already got those headaches so you're not really losing anything. 2. Store the value somewhere and use it to set the field when the popup is displayed. This is one of the reasons I like making each form in a C module of it's own. I can add a module-level variable to the form's C module and expose a setfieldname() function in the header. When the data is read I call the set method to store the value for later use by the form. When the form is popped up it's initialisation function uses the stored value to populate the field. I then store any modified value back in the local variable and use a getfieldname() function to retrieve it at other times. This disconnects the data storage from the form visibility. Of course this uses global variables and requires good modularity to your project so it's not going to suit some people. But then if you're going to create the form and hold it for later use you'll need to store the form pointer somewhere anyway so... It's really a matter of how you prefer to organise your code. My apologies if I've raised more questions than I've answered. I hope at least you've got some alternatives to think about. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: A user that has a full mail box!
From: Luc Le Blanc [EMAIL PROTECTED] This sometimes occurs with people with palmsource.com addresses ;) Regarding the full mailbox problem, shouldn't these messages be sent to the news mailer, not the poster? After all, to the mailer is the message origin (spreading the post), not to poster. I asked about this some years ago but the reply was that it's really out of their control because the mail arrives at that person's inbox with From=WhoeverPostedTheMessage so the mail reader does what it's been told and sends a reply to say that this person is out of the office and can't respond. I've checked the headers and, as far as my limited knowledge takes me, all of the other lists I'm on seem to run the same way. I think this list has this problem more frequently because it's huge and has a fair number of subscribers in corporate environments (where autoresponders are more common). The only way to change this behaviour would be to alter the list manager so that messages where sent using a From=trash so that autoreplies went to a dump address. But if you do that then you lose any reader functionality that depends on getting the author of the message from the From= line. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Trying to test for false and break routine
From: druid [EMAIL PROTECTED]
trying to determine if the field is empty and poping up a
alert window then breaking the routine
been at it a few days now, Im missing something
I think your outer logic is fine, but it's failing because inside
GetFieldData() nofield is set if the text buffer or the field is
missing - these shouldn't happen if your code is written
correctly. What you want is to have that flag set if your
field is *empty*. Change GetFieldData so that the
if (FldGetTextLength (pField))
that does the copy has an else that sets nofield to true.
Your outer logic then has more chance of doing something
useful.
Chris Tutty
(snip)
Boolean nofield = false;
static Boolean frmMain_frmTestButton_OnSelect(EventPtr event)
{
// Insert code for frmTestButton
GetFieldData( fldTest, testdata, 255 );
if (nofield) status = FrmAlert(frmAleart);
nofield = false;
break;
return true;
}
(snip)
Boolean GetFieldData (UInt16 fld, Char *text, UInt16 maxLen)
{
FormPtr pForm = FrmGetActiveForm ();
FieldPtr pField = FrmGetObjectPtr (pForm, FrmGetObjectIndex (pForm,
fld));
if (text != NULL) *text = '\0'; // initialize
if (text == NULL || pField == NULL)
{
nofield = true;
return false;
}
if (FldGetTextLength (pField))StrNCat (text, FldGetTextPtr (pField),
maxLen);
return true;
}
--
For information on using the PalmSource Developer Forums, or to
unsubscribe, please see http://www.palmos.com/dev/support/forums/
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: form - return to previous form
From: Evan [EMAIL PROTECTED] FrmPopupForm the new form then FrmReturnToForm to go back previous form -- Another alternative that sticks to standard FrmGotoForm calls is to store a property with the id of the calling form and use that to return. This isn't perfect, but avoids some issues with FrmPopForm. I honestly can't remember what they were, but I know the popup and return gave me some headaches, possibly one being that you *have* to use the return, you can't then step off to a third form. This can cause your interface to become unnecessarily constrained. Of course, if that's all you'll ever want to do then the constraint isn't an issue. As far as the use of globals goes, I tend to implement each form in it's own module with a header to expose the interface. Properties such as returnToForm are then local to that module and managed using a setReturnToForm() function exposed via the header. While this is still global from the compilers point of view it provides good modularity for code management. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: revisit the alert issue
From: druid [EMAIL PROTECTED]
there has got to be a better way, this could get ugly
with 10 fields or so
You could wrap the test and alert in a private work function
that's passed the name and index of the field. That means
that your main code would just be:
if (handleField(Name, fldName))
{
if (handleField(Number, fldNumber))
{
}
}
That's slightly less efficient, but is a lot easier to read. It also
stays simple even if the logic associated with each field becomes
more complex.
For this sort of problem I've also sometimes used a loop with
a case statement:
for (fieldnum = 0; fieldnum FIELDCOUNT; ++fieldnum)
// common set up logic
switch (fieldnum)
{
case 0: // do stuff for field zero
break;
case 1: // do stuff for field one
break;
// etc...
}
// common error check and clean up logic
if (err != NOERROR)
{
break;
}
} // end field for
which isn't as clean as a worker function but is useful where
the worker function would otherwise need to pass a dozen
parameters in and out. It's essentially just using the loop to
step through a sequence (a simple state machine) to ensure
that the before and after logic in always consistent.
Chris Tutty
#include PalmOS.h
#include testaleart.h
#include testaleart_res.h
Char testdata[256], testdata1[256];
Boolean nofield = false;
UInt16 status;
static Boolean frmMain_frmTestButton_OnSelect(EventPtr event)
{
UInt16 fld;
// Insert code for frmTestButton
GetFieldData( fldTest, testdata, 255 );
if(nofield)
{
status = FrmCustomAlert(frmAleart, Name, NULL, NULL);
nofield = false;
}
else
{
GetFieldData( fldTest1, testdata1, 255 );
}
if(nofield)
{
status = FrmCustomAlert(frmAleart, Number, NULL, NULL);
nofield = false;
}
else
{
file://go to routine to save records
}
return true;
}
Boolean GetFieldData (UInt16 fld, Char *text, UInt16 maxLen)
{
FormPtr pForm = FrmGetActiveForm ();
FieldPtr pField = FrmGetObjectPtr (pForm, FrmGetObjectIndex (pForm,
fld));
if (text != NULL) *text = '\0'; // initialize
if (text == NULL || pField == NULL)return false;
if (FldGetTextLength (pField))
{
StrNCat (text, FldGetTextPtr (pField), maxLen);
}
else
{
nofield = true;
}
return true;
}
--
For information on using the PalmSource Developer Forums, or to
unsubscribe, please see http://www.palmos.com/dev/support/forums/
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: Checking to see if field has data in it before passing it routine
From: druid [EMAIL PROTECTED]
Here's a better way that has error checking, default processing and takes
advantage of a unique feature of the PalmOS StrNCat() function to
prevent
stack corruption:
Are you sure it's StrNCat you want and not StrNCopy? I'm just
asking because the old version copied to text so you seem to
have changed the way the function works.
When using a functin like this where does the maxLen
value come from when passing the data to this function
It depends on how the buffer is defined. If you're defining it
using a constant then pass that constant - 1. It gets more
complicated if you're building a string up and concatenating
because you need to take the existing length of the string into
account but for StrNCopy it's generally just a matter of passing
the defined length -1.
Chris Tutty
Boolean GetFieldData (UInt16 fldNbr, Char *text, UInt16 maxLen) {
FormPtr pForm = FrmGetActiveForm ();
FieldPtr pField = FrmGetObjectPtr (pForm, FrmGetObjectIndex
(pForm, fldNbr));
if (text != NULL) *text = '\0'; // initialize
if (text == NULL || pField == NULL)
{
missing_field = true;
return false; // I like error checking !!
} // use missing_field to set an aleart window and brake the
file://static Boolean frmMain_saveButton_OnSelect(EventPtr event)
// so as to not allow the save ??
if (FldGetTextLength (pField))
StrNCat (text, FldGetTextPtr (pField), maxLen);
return true;
}
--
For information on using the PalmSource Developer Forums, or to
unsubscribe, please see http://www.palmos.com/dev/support/forums/
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: Checking to see if field has data in it before passing it routine
From: Chris Tutty [EMAIL PROTECTED] From: druid [EMAIL PROTECTED] Here's a better way that has error checking, default processing and takes advantage of a unique feature of the PalmOS StrNCat() function to prevent stack corruption: Are you sure it's StrNCat you want and not StrNCopy? I'm just asking because the old version copied to text so you seem to have changed the way the function works. Ah, my mistake, I didn't realise that Roger had stepped in and rewritten your function (of course if I'd actually read the whole function I'd have seen that). And he's added a technique for handling the missing fields. Yeesh, I should just go back to bed. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Checking to see if field has data in it before passing it routine
From: druid [EMAIL PROTECTED]
I need to check to see if the field has data in it before I pass it to
GetFieldData.
What is the best way to do that and assign a text value to name
if the field is NULL
Do you mean if the field is NULL or if the field contents are NULL?
If the contents are NULL GetFieldData will handle it acceptably,
although it would perhaps be better to clear text before copying any
value in fld. Clearing it first gaurantees that there isn't old data in
text after GetFieldData returns.
Unfortunately you're not passing enough information to GetFieldData
for it to gaurantee robust results. All functions that pass a buffer as a
char * must also pass a buffer length. WIthout this length it's impossible
for the function to either clear the buffer if fld is NULL, or to gaurantee
that buffer overflow isn't occuring. StrCopy is actually very dangerous
- your code will be much more robust in the long term if you remove
every StrCopy and replace it with a StrNCopy using the appropriate
defined buffer length AND terminate the string properly. Since the
standard Palm OD StrNCopy doesn't gaurantee string termination it's
better to wrap this in a function that always terminates the string.
It's a bit of work to make sure that you always handle strings robustly,
but it'll save you time in the long run. Tryng to track down buffer
overruns after release is insanely expensive in times of time, money
and product credibility.
So
GetFieldData( fldName, name );
should be GetFieldData( fldName, name, nameLen );
static void GetFieldData( UInt16 fld, Char *text )
{
static void GetFieldData( UInt16 fld, Char *text, length )
{
FormPtr frm = FrmGetActiveForm();
FieldPtrfldP = FrmGetObjectPtr(frm, FrmGetObjectIndex(frm,fld));
MemHandle hText = FldGetTextHandle(fldP);
if (hText)
{
MemPtr *pMem = MemHandleLock( hText );
StrCopy( text, (Char*)pMem );
StrNCopy( text, (Char*)pMem, length );
text[length - 1] = '\0';
MemHandleUnlock( hText );
}
} else
{
MemSet(text, length, '\0');
}
}
Chris Tutty
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: Checking to see if field has data in it before passing it routine
From: Jonathan King [EMAIL PROTECTED]
Here is a simple way to do what you are doing.
static void GetFieldData( UInt16 fld, Char *text )
{
FormPtr frm = FrmGetActiveForm();
FieldPtrfldP = FrmGetObjectPtr(frm,
FrmGetObjectIndex(frm,fld));
if(FldGetTextLegnth(fldP))
which needs to be:
if(fldP FldGetTextLegnth(fldP))
if the point about the field itself being null is to be handled,
but since the code already tests the field handle for NULL
the only other thing you're doing is checking for an empty
string which, from memory, StrCopy will handle acceptably
so I'm not sure that this code adds much and...
{
StrCopy(text, FldGetTextPtr(fldP);
... you're not testing the FldGetTextPtr() return value for
null and, as far as I'm aware, this can return null so your
code is more fragile than the original druid proposed.
No?
-Original Message-
From: [EMAIL PROTECTED] [mailto:bounce-449914-
[EMAIL PROTECTED] On Behalf Of druid
Sent: Friday, July 22, 2005 8:53 PM
To: Palm Developer Forum
Subject: Checking to see if field has data in it before passing it
routine
I need to check to see if the field has data in it before I pass it
to
GetFieldData.
What is the best way to do that and assign a text value to name
if the field is NULL
GetFieldData( fldName, name );
static void GetFieldData( UInt16 fld, Char *text )
{
FormPtr frm = FrmGetActiveForm();
FieldPtrfldP = FrmGetObjectPtr(frm,
FrmGetObjectIndex(frm,fld));
MemHandle hText = FldGetTextHandle(fldP);
if (hText)
{
MemPtr *pMem = MemHandleLock( hText );
StrCopy( text, (Char*)pMem );
MemHandleUnlock( hText );
}
}
--
For information on using the PalmSource Developer Forums, or to
unsubscribe, please see http://www.palmos.com/dev/support/forums/
--
For information on using the PalmSource Developer Forums, or to
unsubscribe, please see http://www.palmos.com/dev/support/forums/
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: Checking to see if field has data in it before passing it routine
From: druid [EMAIL PROTECTED] The problem is I need to know what is in fldName before I try and use GetFieldData and if the contents or the field is NULL not do the call to the GetFieldData routine Nope, I think you're doing the right thing bu checking for NULL data within the function. This means that the function is safe regardless of what's passed to it. You just need to check the fldP for null and I think the function as written is fine. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Checking to see if field has data in it before passing it routine
From: druid [EMAIL PROTECTED] I tried you code change and it does work Now for the wierd thing I added this StrCopy(text, UNKNOWN); } now if I dont fill in any of the fields and it writes this to the record, the very first field has the first three characters cut off OWN,UNKNOWN,Select,, Hmm, that's the first four characters which is significant because it's the size of a long int, suggesting that a char pointer has been offset by one (stack corruption, mishandled pointer or something strange about the way the constant is being stored). It's difficult to say why this might work for good data and fail for the constant. Try checking to see if the data *before* the first field has been overwritten (generally requiring some work with the debugger or a hex editor). You want to know whether the full text has been written to an address four bytes before the start of the field or whether the start of the constant is wrong by four bytes but the data has been written to the wrong location. There are some compiler settings for how constants are handled but I wouldn't have thought they'd cause this sort of problem. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Palm dedicated to one application
From: Jerry Martinez [EMAIL PROTECTED] We have customer who wants his palm to only run our palm app when he turns on his palm and does not want access to any other feature. Is is possible? If the client has the budget one alternative is to create a special ROM image that only includes your app and drops the standard Palm OS apps. This generally isn't something you'd do for one person but I mention it because it's an often-overlooked option for the distribution of custom-purpose units within an enterprise or as prizes or advertising hand-outs. That takes a little preparation, but I was pleasantly surprised by how easy it was - of course this requires devices that have flashable ROM. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: im stumped could use some ideas
From: druid [EMAIL PROTECTED] I run my program and get no errors However looking at the records with par -d-- 0 68 New QUest.22.Fair.123456789.good.A Gnoll Cave.NO I always end at 68 to 70 characters From memory this is a par problem - I think the output was only intended to dump database contents for debugging so they never bothered writing code to handle any string size. If you want to handle string contents you need to move to a different tool - there are a couple that convert databases to csv or text files that handle simple databases quite well, or a commercial tool like PDBGo that will let you write custom scripts to extract database contents to exact specifications (there also used to be a PDBC tool that allowed COM access to PDB files from memory). Google should answer this. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Ideas on how to do this
From: druid [EMAIL PROTECTED] I have to many fields for one form So I want to make a popup form for some of the checkbox fields I was going to use a FrmPopupForm call then a FrmReturnToForm(0) call to go back to the origional form. This can cause problems if it's not done carefully. You should review the archives of this forum for more info. I guess I could make global variables, but is their a better way to do this. If you mean having a global structure that stores data shared by fields spread across several forms then I think this has the solid advantage of making it much easier to move fields between forms or add other forms as you extend the project. This is true regardless of whether you use a popup or not - you want to avoid having the data management for the app hard-coded to the fields that happen to be on each form. If your app is getting to this size it's time (past time?) to create a code module to manage the data separately from the form code. This module would have the code to load and save the data from the database and would either expose globally visible variables or, more work but better, expose get~ and set~ methods for each property. Just because you're coding in C doesn't mean that you can't use the object-oriented concepts of data hiding and interfaces. They're really just old C concept anyway. I guess it's obvious advice, but a great deal of the simple examples for Palm have the field management logic in the same functions that implement the forms so it's easy to extend these simple examples without stopping to think that at some point their underlying design has to be re-examined. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Why does this cause a low byte fault solved update I think
From: druid [EMAIL PROTECTED]
This is what worked finaly
Thanks to all who helped, it seems that when you create
any key or enter event, all default action is ceased
and must be hard coded.
Well, as I indicated in my email, when I used
static Boolean frmMain_fldName_OnfldEnter(EventPtr event)
{
SysKeyboardDialog(0);
return true;
}
it worked as expected and didn't fault. I only tried four or
five iterations but this seemed sufficient since you'd reported
a crash after three. I still suspect that there was something
else going on with your code. Your last response that when
you took out the keyboard popup the crash went away doesn't
mean the rest of your code is bug-free, just that the problem
went away. This can sometimes mean that memory or stack
corruption is happening, but that's it's not affecting important
areas of memory. You can also have code that will work OK
as long as you don't change the active window.
It's good that you can continue moving forwards, but I'd treat
the underlying problem as mising rather than fixed.
Chris Tutty
UInt16 entered_name = 0;
static Boolean frmMain_fldName_OnfldEnter(EventPtr event)
{
// Insert code for fldName
FormPtr form;
Boolean handled = false;
FieldPtr fldP;
form = FrmGetActiveForm();
FrmSetFocus(form, FrmGetObjectIndex(form, fldName));
fldP = FrmGetObjectPtr(form, FrmGetObjectIndex(form, fldName));
if(entered_name 1) EvtEnqueueKey(vchrKeyboardNumeric, 0,
commandKeyMask);
entered_name = 1;
FldDrawField(fldP);
handled = true;
return true;
}
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: Crash if I click button twice
Druid,
It looks fine at first glance.
My only comments right now would be:
- after labela = (Char *)CtlGetLabel( ctlP ); you should check labela
against null before using it. From memory it's possible for a popup
to have nothing selected in some circumstances.
- How does the GetFieldData() know to limit the data to 256? I would
have thought that this function would also need to be passed a
buffer length.
Aside from that it all looks sensible.
Chris Tutty
- Original Message -
From: druid [EMAIL PROTECTED]
To: Palm Developer Forum palm-dev-forum@news.palmos.com
Sent: Tuesday, July 19, 2005 11:34 AM
Subject: Crash if I click button twice
this code will crash If I click the save button twice
I must not be releasing some memory somewhere
static Boolean frmMain_saveButton_OnSelect(EventPtr event)
{
Char name[256];
Char level[256];
Char expFld[256];
Char desc[256];
FormType *frmP;
ControlType *ctlP;
Char *labela;
Char labelb[256];
frmP = FrmGetActiveForm();
ctlP = FrmGetObjectPtr( frmP, FrmGetObjectIndex( frmP, frmPopupTrig1 ) );
labela = (Char *)CtlGetLabel( ctlP );
StrCopy(expFld, (const char *)labela);
if(dbPtr == NULL)
{
// First see if we can find it:
LocalID dbID = DmFindDatabase(DB_CARDNO, DB_NAME);
if(!dbID)
{
// Couldn't find it, so create it:
Err err = DmCreateDatabase(DB_CARDNO, DB_NAME, DB_CREATOR, DB_TYPE,
false);
if(err)return err;
// Now we ought to find it:
dbID = DmFindDatabase(DB_CARDNO, DB_NAME);
if(!dbID)return 1;
}
// Found it, now open it:
dbPtr = DmOpenDatabase(DB_CARDNO, dbID, dmModeReadWrite);
if(dbPtr == NULL)return 1;
}
// Everything went smoothly:
GetFieldData( fldName, name );
GetFieldData( fldLevel, level );
GetFieldData( fldNewDesc, desc );
size = StrLen(name)+1+ StrLen(level)+1+ StrLen(expFld)+1+
StrLen(desc)+1;
index = dmMaxRecordIndex;
h = DmNewRecord(dbPtr, index, size);
ptr = MemHandleLock(h);
Offset = 0;
DmWrite(ptr, Offset, (const void *)name, StrLen(name)+1);
Offset += StrLen(name)+1;
DmWrite(ptr, Offset, (const void *)level, StrLen(level)+1);
Offset += StrLen(level)+1;
DmWrite(ptr, Offset, (const void *)expFld, StrLen(expFld)+1);
Offset += StrLen(expFld)+1;
DmWrite(ptr, Offset, (const void *)desc, StrLen(desc)+1);
MemHandleUnlock(h);
DmReleaseRecord(dbPtr, index, true);
err = DmCloseDatabase(dbPtr);
return true;
}
--
For information on using the PalmSource Developer Forums, or to
unsubscribe, please see http://www.palmos.com/dev/support/forums/
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: Crash if I click button twice
From: druid [EMAIL PROTECTED]
this code will crash If I click the save button twice
I must not be releasing some memory somewhere
dbPtr seems to be external. You're closing this database each time, but
keeping the dbPtr and trying to use it next time.
You need to be clearer about who owns dbPtr - if it's owned externally
then the db find, open and close should be external, if the open and
close are going to be done by this function then it needs to own the
dbPtr.
A fast fix is to set dbPtr to null after the close.
Chris Tutty
static Boolean frmMain_saveButton_OnSelect(EventPtr event)
{
(snip - no declaration of dbPtr)
if(dbPtr == NULL)
{
(snip of find, create and open db)
}
// Everything went smoothly:
(snip of create record and write data)
err = DmCloseDatabase(dbPtr);
return true;
}
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: windows console application
From: Mike Hodkin [EMAIL PROTECTED] Need to walk before running. I need to brush up on C/C++ before diving into Palm. The Palm stationery produces apps that will run (from memory a simple form with and edit field and an About form). You can use this as a platform to practice C skills just by writing background modules that calculate or manipulate strings and then copy the results to one of the pre-built forms. I had a student that wrote code to calculate student debt repayments and produced the output by using the window write routines to dump text to the current form. You don't have to know anything about the Palm OS memory management, form, controls, events or database functions to do this, but can gradually expand your tests to include those. The problem I've found with students working on C skills outside of Palm OS is that it's difficult to ignore the environment issues, particularly when the Palm libraries don't use the standard C naming for core functions (what-ever misguided Pascal-oriented idiot decided to rename standard C functions deserves a good kicking but it's with us for good now). Although you will run into annoying walls with Palm OS issues this group should be able to fix those for you fairly quickly. At least with that approach you get your head into the right space in terms of the design, code, debug cycle. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Using the Table Control
From: Layne Lund [EMAIL PROTECTED] values are all static in the resource file. Is there a way to set the column widths at run time instead? Of course, this gets into a whole other mess in trying to calculate the width of a string in a given font, but I can figure out that part later. I need to first know if it's even possible to set the width at run time. (Maybe I should just That's actually the easy part. There are a number of good functions for determining pixel widths and wrap breaks of text. Example: // set drawing font FntSetFont(g_prefs.Font); LstGlueSetFont(GetObjectPtrFromId(frmP, MsgListMessageList), g_prefs.Font); // Calculate widest column contents (assumes fixed width digits) LineOneX = FntCharsWidth(00/00, 5); I then draw the whole list myself using. // Notify Palm OS that we will draw each list item ourselves. pList = GetObjectPtrFromId(frmP, MsgListMessageList); LstSetDrawFunction(pList, ListDrawFunc); and combinations of WinDrawChars() and WinDrawLine(LineOneX + 1, etc to do the work. It's a bit fiddly to set up, but gives you complete freedom to change the style of the list and my experience is that once you're concerned with column widths you're likely to start wanting to make all sorts of detailed changes. Biting the bullet to go to custom drawing pays off when you're fine-tuning the list style. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Why does this cause a low byte fault
From: druid [EMAIL PROTECTED] testdb just read from memory location 0x324A which is in Memory Manager data structures Boolean handled = false; form = FrmGetActiveForm(); FrmSetFocus(form, FrmGetObjectIndex(form, fldNewDesc)); fldP = FrmGetObjectPtr(form, FrmGetObjectIndex(form, fldNewDesc)); crashes here on the third time this is called - SysKeyboardDialog(0); FldDrawField(fldP); -- You're not testing return values at all so if either the FrmGetObjectIndex() or FrmGetObjectPtr() calls fail (say because the resource files haven't linked properly or fldNewDesc doesn't exist on the current form) then all you'll get is a crash. If you test fldP for null before trying to use it your code won't crash - of course it won't do what you want, but at least you'll have more chance of seeing what the problem is. However, that's probably an aside since a crash the third time you're doing something suggests that by the time you get here something in the first two interations has corrupted memory or the heap. The problem might not be in this code at all. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: freelance help / active posters
From: Henk Jonas [EMAIL PROTECTED] Aaron Ardiri wrote: Ben has worked for Palm Source and now works for Palm, Inc. he was an active posted before he started working there - but, i do remember the days when i used to hold the 'most active' post crown :) 2938 messages http://www.escribe.com/computing/pcpqa/index.html?by=Authora=Aaron% 20Ardiri 5100 messages http://www.escribe.com/computing/pcpqa/index.html?by=Authora=Ben% 20Combee I only get 623 posts for me - no stop, it's 624 :-) Well yeah, and I get 719 (heh - never thought of using the archive for that) but if the archive totalled 'questions' and 'useful answers' separately the contribution of people like Ben and Aaron would really become clear. Which is not to say that they're alone, one of the things that's made Palm OS development easier for me over the years is the extent to which this group will help with any question, no matter how trivial. Many forums that are able to answer specialist questions become quite short-tempered with newbies. It's the fact that this group has always had people with the patience to answer the simple questions that's helped to keep it strong (group hug? Come on, don't be shy). And apologies to druid for starting a side-conversation without re-subjecting his thread. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: freelance help
From: druid [EMAIL PROTECTED] anyone know where I can find some part time assistance for hire I find I am getting into areas that my six books do not explain well. Learning by doing is working except I spend hours trying to figure out something that is simple, and time is money so in the PALM world im going broke LOL This group is also good for reasonably fast turn-around on questions so I think even if you do retain someone to assist it's always worth searching the forums or posting a question here first. I'm constantly amazed at how quickly obscure questions are answered here and for the simple questions you'll often get two answers and a code fragment. Of course you have to walk away from the problem for a few hours and work on something else but then that's not always a bad thing anyway. It's generally where you're the first person trying to do something that the group can't help. As an example, I think Vesselin has answered about ten times as many questions for other people as he's had answered of his. Actually it's about time PalmSource put up a monthly cash prize. I know there's about a dozen people I'd vote for and it's got to be saving PalmSource money because without this group they'd have to provide real support :-) Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Tungsten W emulator mangles bitmaps.
From: Valeriy Zamarayev (DHL UA) [EMAIL PROTECTED] The problem appeared with Tungsten W emulator. This is I believe the only OS4 handheld which supports the high density feature. The problem is that during animation images are often mangled after some period of play, they are a kind of rotated, the sprites are shifted to the right, and pixels from the right border appear at the left border of the sprite. It's quite random, because sometimes bitmaps are drawn correctly, sometimes not. How much is it shifted by? Could it be an issue to do with rounding the bit image to a byte or multi-byte boundary? Is the shifting only once for the image or for each scan-line? I haven't done low-level image stuff for ages, but I recall that some shifting occurs because the image block is mis-aligned while some is because the image isn't the 'right' width and each scan line will be affected. If the shifting gets worse down the image it's a per scan-line thing where-as if it's constant down the image it's more likely that the start of the image data block is somehow incorrect. Anyone that's actually done this stuff under Palm OS can feel free to step in and correct me (Aaron? hello? Paging the GamesMeister...). Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: -1 madness?
From: Bryce Burrows [EMAIL PROTECTED]
Using PODS , i have a function i return the int value -1 from
int MyGroovyFunction()
{
int nMyReturnValue =0;
// do some funky work and for a very good reason set the following
nMyReturnValue = -1;
return nMyReturnValue;
}
While I agree that this should work I walked away from
using -1 as an error return a while ago because of this sort
of problem.
If you use error enums or constants it doesn't matter what
the number is and you avoid signed vs unsigned issues
completely. This makes it more difficult to pass back a
value that's either data if it's = 0 _or_ 'undefined
error' if it's 0 but then experience showed that there was
more potential for obscure bugs with this sort of value
combination than by returning an error code and using
another method to obtain a valid value. I just got sick of
finding that an error value was being used as valid data
because someone forgot that their (if x 0) was using an
unsigned x (not that you are) rather than a signed value.
Alternatively your function only returns a valid value or zero
and use another mechanism to identify a problem. I've got a
number of modules where I've implemented a xxxGetLastError()
function. The code logic then tests this after each call rather and,
if it's NoError, uses the return value from the function. It's a
mechanism that's far more robust and extendable and if you
can get used to implementing it from day one it's no slower
to write.
I guess to a certain extent I use these methods because I
code in a team with junior programmers and I've got more
need to protect against poor judgement than one experienced
coder working alone, but then I think my coding style has
improved because of the need to establish more formal error
handling mechanisms.
Feel free to flame me for issuing a condescending lecture on
code style rather than working out what the problem is. :-)
Chris Tutty
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: -1 madness?
From: Bryce Burrows [EMAIL PROTECTED]
ahh something seriously screwy is going on with PODS
if (false == bFailed)
{
nCount ++;
}
bFailed is true , yet it heads on to the nCount++ line
I've found that with compiler optimisations small test code
often doesn't operate as you would expect because the
compiler eliminates code without side-effects or merges
code paths that resolve to the same end result.
What I mean is that if the code after the if () continues
on to a code section that also executes nCount ++; the
compiler might determine that the test is redundant. The
debugger is only approxiamating the connection between
the compiled binary and the source text and so can
sometimes indicate that a certain line is executing where
what's actually happening is that a line with the same effect
and sequence in the binary code flow is executing (if that
makes sense at all). When you add register colouring to the
equation it gets even more complicated.
My suggestion is to add an else clause that does something
different whichcauses sufficient side-effect that the compiler
is forced to execute it and see what the debugger shows
when you test that.
Chris Tutty
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: freelance help
From: druid [EMAIL PROTECTED] anyone know where I can find some part time assistance for hire I find I am getting into areas that my six books do not explain well. Learning by doing is working except I spend hours trying to figure out something that is simple, and time is money so in the PALM world im going broke LOL I've trained several Palm OS developers and have mentored people working through self-teach programs via email. If you're interested contact me off-list and we can discuss rates and structure. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: libexpat?
From: Scott Bennett [EMAIL PROTECTED] Just wondering if anyone has tried porting libexpat to palm (its an XML library), If you did attempt it, what sort of problems did you encounter? I've seen a couple attempts at XML parsers on palm but they all have their problems, I even found a commercial one that had a 64k limit to document size (I'm guessing they dont use the glueptrs). -- From memory the problem with porting most desktop XML code is that it handles the unknown depth of the hierarchy using stack-based mechanisms. This produces code that is fragile under Palm OS because of the smaller stack. You generally have to refactor the code to use memory-based mechanisms for building and iterating the tree, making the job more complex than a simple port. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: ptr is handle error?
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] This is definitely a case of accessing a non existent memory location. Ah, are you sure about that? I thought that accessing non-existent memory tends to produce other kinds of errors - NULL pointer references, access to unallocated memory, etc. But ptr is a handle? It sounds as the program is using a MemPtr (or some other pointer) at a place where a MemHandle is expected (or maybe the other way around?). But I thought that the compiler would catch such things... On the few occassions that I've had to chase memory contents around in hex it seemed like the hex encoding of handles was deliberately outside of the valid values for pointers. I think Palm OS is built to check for a pointer that 'looks like' a handle. I don't think the compiler can always tell you if you've accidentally addressed a handle as a pointer (I don't remember what MemHandle and MemPtr evaluate to, but I've had this happen before when not being careful enough with casting of return values). I've also got a feeling that the older types were more prone to confusion but I'd have to dig a project out of storage to be sure. If I'm correct about the logic being based on the the value of the pointer, this could be confused by memory overwrites caused by wild pointers or heap corruption, although I think it's more likely that somewhere you're not being strict enough with casting or variable typing. This is nasty to track down blind so it's worth putting some work into trying to reproduce it. See if you can get a copy of the emulator session your client is working with, or any details as to the sequence of events that cause it. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Re-arranging functions in a code segment
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] The code of my application is signifficantly less than 64 Kb - it is only about 45 Kb. Nevertheless, long time ago I started getting text segment full errors from the linker of PODS (ld). Obviously, this happened because the distance between some function and the function it was calling had become larger than 32 Kb. (snip) My question is - is there some clever way of doing this? Alas, the linker doesn't tell me which particular function was trying to call a function that was more than 32 Kb away, so I didn't know which function to try moving closer to which one. That's a pity - it looks like PODS is significantly less useful than CW in this respect since this wort of problem in CW would produce a 'function out of range' error or some-such and identify the module attempting the call and the function being called. That made it fairly easy to pin down the problem. I never had significant problems running segments of 40K to 50K, but then generally there were a couple of core modules that called out to secondary modules and putting those in the centre of the segment had a good chance of making all functions within reach. With apps that have more complex inter-connectivity it can get complicated. I'm not sure how easy it is to compile a PODS project under CW, but that might give you the info you need to make the app single-segment. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Object Library for Palm OS (was Re: Old SDKs)
From: Ben Combee [EMAIL PROTECTED] At 10:05 PM 6/26/2005, you wrote: Are there any plans to port POL to prc-tools so that it can be used with the new Palm Development Suite? Personally, I like the whole object-oriented paradigm. From your brief explanation here, it sounds like OO provides a great way to hide the differences in the API changes. Not that I know. In my opinion, the C++ support in prc-tools isn't quite able to handle POL very well -- there's no exception handling support for 68K code, something that POL uses. It's also more difficult to partition code into sections, something that POL uses in its static library setup to be able to pull a lot of code into a program. The author did look at a prc-tools version a few years ago, but decided not to go ahead with it at the time. He's the one who has rights to any future versions of the library. It's a pity that POL is so promising and still so constrained. I looked into publishing some of our routines as a C library for abstracting the version issues but both the Glue layer and POL seem to solve half the problem, but different halves (Glue is good for some problems, but not others, while POL seems to be good for more problems, but only if you're coding in C++). I'd like to see that effort aimed at creating something that is useful to a wider range of Palm OS coders, hell even the initial default projects could be much more useful, but the framework needed to allow a range of programmers to contribute is too big a job for anyone to take on as a hobby. It's a pity and, IMHO, the sort of thing that stays unresolved unless someone puts up a stack of money to fund the framework and, as for IBM and eclipse, there's got to be a good business reason to do that. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Visual FoxPro/Palm Custom Conduit Help
From: Dan Wellisch [EMAIL PROTECTED] I want to develop a conduit application that can sync .pdb files on the Palm with Visual FoxPro .dbf files. I haven't done this, but I worked with complex integration with an Access app some years ago and have some alternatives you could consider. 1) Hotsync from PDA to Desktop 2) Run a .pdb to .dbf converter (avail. from a 3rd party) 3) Write a Java app to read the converted .pdb in 2) and process that one against the .dbf in the Visual FoxPro application. In essence, I would be creating the hotsync to happen using the .dbf format (via ODBC). I know this works. 4) Run a .dbf to .pdb converter (avail. from a 3rd party) and replace the .pdb in the Desktop Hotsync dir with the updated .pdb from process in 3) There is a commercial app called PDBGo that could handle 2), 3) and 4) from one application. It's a COM component that comes with sample code in multiple languages to read and write pdb files. It's also worth remembering that record-based Hotsync isn't the only way data comes off a device. We found it easier to work with the database backups HotSync stored in the user backup folder. We picked these up intact, carried out a data import to the Access database using PDBGo and then created new databases for installation to the device. In our case the app was a data collection app using Access- -generated lookup tables so the downloaded database was a complete replacement for the device database. If you want to syncronise the device database it's a little more complex. Another approach I've used successfully with other projects is to build throw-away update pdbs. By that I mean a small pdb with new records in it. On startup the app looks for databases having the 'update' naming pattern (and, of course, the correct CRID). If one is found it uses the contents to update it's primary database and deletes the update pdb. Certainly a strong driver in developing these solutions were deficiencies in the earlier HotSync and conduit solutions available and things have changed significantly in the conduit area since then. As an example I've seen examples in this group of Delphi-based conduits that seemed to do a great deal of work fairly easily. So I'm not suggesting that these ideas are the best way to do things, just some alternatives you might keep in mind. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Perl scripts for POSE
From: epross [EMAIL PROTECTED] I seem to recall seeing a few scripts that had the functionality to automate POSE and take screenshots, etc. Does anyone know where I can find these? They used to be distributed with the emulator download, in a scripting folder. I'm not sure if the modern downloads include them. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: New guy questions
Aaron D. Campbell wrote: I have a few questions. All are directed at development for the Treo 650. 1) Can GPS location be integrated into an application, and does it require additional hardware? I want to be able to send current coordinates to my program. 2) Is it possible to allow someone to write on the screen? I want to get a signature from a delivery location. 3) Can an app send post requests to a website, and receive data back? 4) How hard is it to integrate barcode scanning into an application (preferrably a bluetooth scanner) 5) I'm a web programmer, who has no experience developing anything but websites for the palms. Where is a good place to go to find a developer to create this app? It sounds like you're building yet another courier support application. I'd write up a spec detailing your requirements and then announce the project here requesting bids. I'm sure that this has been done half a dozen times before so another approach might be to search PalmGear and google for existing products. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Old SDKs
From: Tam Hanna [EMAIL PROTECTED] I don't beleive that artificially restricting to OS5 is a good idea. Just think about all the Zires and the existing user base. I always try to target OS3.5, as that is a pretty good common denominator(first 8meg devices,...)! The difficulties with pre-OS4.1 are you get into times when the base APIs were still in a state of flux. This affects things such as form painting (the event sequences and the way background painting was done changed), the serial libraries changed, sound management changed (I think?). Not that I want to imply that this has stopped happening. :-) Our code for 3.1 to 4.1 support effectively has an abstraction layer that works out what routines are appropriate for the OS version and adjusts. It's a pain to support and debug. While I agree that you can often support a wider variety of devices by coding carefully, it's easy to forget how much has been added to Palm OS. Image buttons, for instance, are pre-4.1 from memory and handling those yourself is just a pain. They're also an example of something that influences the look of the app and can only be worked around by having two sets of resources. The other approach is to make an assessment as to how much of the coding work is interface-driven. For an app that has a lot of code related to storing data, performing calculations and communicating with a server it can be a relatively small task to define a function interface between the core modules (which will change very little over OS versions) and the interface (which will look completely different under a colour 320x320 OS5 device than on a mono 160x160 OS 3.5 device) and then build two interfaces. For your friend build the OS 3.5 interface but tinker with a prettier colour interface that you can flesh out if the app becomes viable as a commercial product. It sounds more complicated to have two interfaces but it's often less work to have two simple, tightly targeted code modules than one complex, multi-version, multi-format code module. If you plan for this from day one you can save some pain downstream. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: PluggedIn Site
From: epross [EMAIL PROTECTED] Thanks for the quick response. Just to note, I only get this with Netscape. With IE it works just fine. What browser are you using? It's showing a problem that has something to do with cookie handling - do you have cookies turned off? Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: giving error as app1 just read from memory location 0*000000001,which is in low memory
From: rupesh kumar [EMAIL PROTECTED] i am trying to access the contacts database.my application name is app1. its compiled well but at runtime when tring to get the data it is giving error as app1 just read from memory location 0*1,which is in low memory. why this error is coming ? In general this means that you're passing a null pointer to an API call, or using a null pointer as a structure reference. The only strange thing is that the value reported is often the offset of the value requested from the start of the struct which should really be an even number because of structure padding. Anyway. The first thing you need to determine is where in your program this is happening, either by debugging or by writing to a log as your program operates. Without this information it is very difficult to identify which value is incorrect. You could also work carefully through your code making sure that every pointer is tested for null. This is time-consuming but is useful to ensure that your code is robust. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: HTML
From: Jerome Chapdelaine [EMAIL PROTECTED] A while ago I looked at the help widget of FLTK. It can render basic html and it would be very easy to port (only the font handling + box rendering if you don't need bitmaps). But it would probably not handle full web pages very well... I would use it only for formatted text. I have a couple of clients and some potential projects that require simple HTML display and would be willing to contribute to a project such as the port described. The only difference is that I'd want to include bitmap display. I don't have the time to take the whole project on. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: HTML
From: James Fisher [EMAIL PROTECTED] Why don't you use the Addit viewer? It renders html passed through a simple content maker. Addit is shipping on the ROM of all new palmOne devices. You can call the viewer with appropriate content and include various buttons, links etc to launch back to your app. It has a ton of functionality available. It isn't free for commercial use but is free in most cases for developers using it in their own commercial apps. Their would be a license fee if you developed the app for a commercial end user. Which is all of the apps I'd use it for. I couldn't find any info on using AddIt in this way - all documentation assumes that it is downloaded for use as a software portal and so doesn't talk about distribution or licensing for other uses. Can you point me to the documents that cover the legal and pricing details? Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: White screen with Treo 650?
From: Paul Jensen [EMAIL PROTECTED] I distribute a couple of freeware PalmOS apps (mainly as a hobby), and I have received a couple of reports from users with Treo 650s. They claim when they run the apps they get a white screen (or in one case that they can see buttons, but not a drawn image). I've seen that before - now if only I could remember what caused it :-) I haven't worked on a Treo650 and I've accidentally caused problems like that so it might not be the Treo. From memory one of the causes was some old code built to allow custom drawing onto Palm OS 3.5 and 4 screens that was messing up under Palm OS 5 and, also from memory, it was related to changes in the event sequences or timing of events sent when a form is opened and drawn. I'm not sure how the speed of the Treo650 relates to the T/E so can't comment on whether this might be affecting event sequences (leading to the problem not presenting for you). If one of the users is willing to put in the time as a tester I'd suggest cleaning up the form open and draw logic specifically for Palm OS 5 and see if that improves things. Also try to make sure that you're not assuming too much about the event sequence you will receive when a form opens. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Treo 650 question
From: Mikael Esesien [EMAIL PROTECTED] I have customer in Germany who can't get my app to run on his Treo 650. He has it installed on the extension card and keeps receiving the following message: Nicht gengend Telefonspeicher frei, um die Anwendung von der Karte zu starten, which he says tranlates as:Not enough telefon storage avalable to start the application from the card. I asked whether or not he meant the SIM card - though he didn't answer. I explained that the app should be copied to the RAM at runtime and not the 'telephone card?' This is baffling me. I have no experience of the 650, so I can only guess what is going on. Is it possible that he's just filled the device up? You could ask him to report the amount of free space on the device. Running your app on a 650 simulator might tell you if the bloat occurs and would help point your client in the direction of the information you want regarding free space. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: A question, v.valev
From: Roger Stringer [EMAIL PROTECTED] It also looks like, based on the couple of Palm OS for Linux sessions I went to at the Developer Conference this week and some private conversations with some technical people from licensees, that PalmSource's implementation approach for Palm OS for Linux looks very good. Their approach should ensure excellent forward compatibility for applications developed for both PACE (68K emulation) and also Protein (native mode) environments. Because of the Protein compatibility of Palm OS for Linux , licensees who implement Cobalt-based devices shouldn't be left out of the application development stream, ... And, more importantly, we might actually get a Palm OS platform that doesn't feel ten years behind the times (64K barriers, no threading, difficulties for C++, etc). When the Palms first came out I was amazed that the power of my old Atari1040 had been squeezed into a 'top-pocket' device but that was a while ago now and the OS is really starting to show its age. What I'm more interested in is whether the Linux kernel will be accessible or whether it will be locked away the same way the original threaded kernel was. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to make a NOT-EVEN length struct?
From: Logan Shaw [EMAIL PROTECTED]
Chrix Bell wrote:
I have make a struct with two offsets and a take-place field
looks like this:
struct whatever{
unsigned Aoffset;
unsigned Boffset;
char FirstField;
}
I want to use this to pack the record size,
What does pack the record size mean?
Since the original message described writing two fields my guess
would be to pack string data from multiple fields into a database
record without unnecessary padding. This is hardly a new concept
for Palm OS databases :-)
I tend to create the struct ending with a char[2] to indicate that this
is string data rather than an individual char and, more importantly,
comment the struct to indicate to any programmer that follows you
that the record data entends past the end of the struct.
data would be write from
FirstField, write A, then from (FirstField+ Boffset), write B.
What are A and B? Are the some sort of data you want to write
into a database? Are they variables? Are they instances of
struct whatever?
Doesn't matter. The concepts are the same regardless of the
objects - find the size of the first and second, create a buffer
big enough for the data and any management overhead, write
the first record to the buffer at offset zero (or just past any record
header), find the size of the second and write at an offset just
past the end of the first block. Then (if you want to use the technique
this poster seems to be following) write the offset for the second
record to the header at the start of the buffer and close the record.
The problem is that you don't write B at (FirstField+ Boffset).
Generally each field is written at the stored offset matching it
so A would be stored at Aoffset (strictly redundant because
you know it starts at the offset of the FirstField struct member),
then Boffset is calculated as Aoffset + (sizeof(A) or StrLen(A))
+ 1 (depending on whether you're storing a struct or a string and
a nasty trap for beginnners) and B is stored at Boffset. Continue
as required. The calculation mistake is probably why you're
getting an out of bound error.
This has been done before so try to find some code that
implements this and copy it.
I am wondering if there are some other kind of skills can make
the record's length not a fixed one.
If by the record you mean the struct, then it is impossible.
In C, a struct's size is determined at compile time.
Strictly correct, but slightly misleading...
If you need
to write a data structure whose length differs depending on what it
contains, you will need to write it by hand, one byte at a time.
(Or with DmWrite() or MemMove() or similar.)
I've seen this technique of a using a struct to describe the fixed
elements and appending strings to the end of it fairly often in
Palm OS apps - I'm sure at least one of the sample apps does
things this way so it's not only not impossible, it's fairly standard
for Palm OS. The confusion is that you've got to understand
that the struct is just a guide that simplifies accessing the start
of the record - the actual data extends beyond it so, as you say,
you've got to be careful about deciding whether you're going to
write the struct as a block and then append data or write the
contents of the struct member by member (generally safer).
If you're only storing strings you actually don't need the offsets
as long as you make sure that each string is null-terminated in the buffer.
You can then read the first string at offset zero and the second
at offset StrLen(firststring) + 1).
Storing the offsets can still be useful if you're writing many strings
and want to quickly extract one without reading the preceeding
strings.
Chris Tutty
--
For information on using the PalmSource Developer Forums, or to unsubscribe,
please see http://www.palmos.com/dev/support/forums/
Re: How to develop an application to support other than english language?
From: babbu cathy [EMAIL PROTECTED] Is it possible to create an application in Palm OS for supporting of some other languages(Japanese or German or French,...) than English? There are a couple of approaches. Palm OS supports localisation (see the Palm OS Companion section on localisation). This basicly works by copying your resources, modifying them for other languages and tagging them with the language they are for. This is all you need and is quite straight-forward for the european languages - I'm not sure if the localisation supports asian languages, but even if it does just modifying the resources and writing your code becomes more complex because you move into multi-byte characters. This approach interacts automatically with the locale in the Palm device to show the language the user has chosen but, although this is handy, means that it isn't always the best answer. I've had clients who wanted their customers to be able to change languages on the fly and for that you'll have to manage langauge selection and possibly resource loading on your own. I wrote a simple mechanism that stored multiple copies of the bitmaps and string resources and loaded the appropriate one depending on the language the user selected. The other problem with localisation that this overcomes is that localisation turns your app into two files - the prc and a set of locale overlays. This works fine if it's all in ROM (which is how PalmSource distribute their apps) but can be a hassle for downloaded apps where end users will be doing the install. There are also commercial localisation frameworks. These are powerful and expensive. If you can afford them they will save your project money in the long term but there can be quite a financial hit to pick them up. As an aside - the SDK used to have a spreadsheet buried somewhere wityh european translations of the standard Palm OS text. Useful if you're doing internationalisation yourself on the cheap. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: displaying a text string that is longer then the screen is wide
From: Jerome Chapdelaine [EMAIL PROTECTED] Use FrmCustomAlert with the ^1, ^2 and ^3 replacement placeholders. -or- Build a custom modal dialog box with a multi-line read-only text field. The dotted underline of text fields can also be removed in the resource editor. If you want to resize your dialog box to fit, you can compute the field height required with FldCalcFieldHeight, resize the controles, then draw your form. etc... There are also text functions that will tell you how much text will fit on a line at the current font size, screen resolution, etc so if the answers above aren't usable it isn't that hard to write a function that wraps text to the screen. Come back if you need more detail on this option. Chris Tutty -- For information on using the PalmSource Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: MemPtrResize
From: Orasanu Iuliana [EMAIL PROTECTED] I'm finding very difficult to work with memory allocation on this platform... :-( I'm trying to reallocate a memory chunk, and my code looks like this: pDBName = (char**)MemPtrNew( sizeof( char*)*2); // i'm allocating memory for only 2 fields, Nope, this problem is reported fairly regularly and there are some excellent detailed explanations in the forum archive but the short answer is that a char * isn't a field, it's a pointer to a field. You've allocated 8 bytes, enough to store two pointers, but what you *really* want is a block of memory large enough to store your two strings (however large they may be). You can either use a fixed-size strategy, and allocate 20 chars, or 40 , or 200, and then limit your strings to that size (less one for the null terminator), or use a packed approach and allocate enought memory for the StrLen(field) + 1, twice. Sorry if this is confusing but, as I say, it's been covered very well just recently by someone. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: MemPtrResize (oops)
My apologies, I didn't read your message carefully enough (thanks to Vesselin for answering properly). Please ignore my earlier reponse. From: Orasanu Iuliana [EMAIL PROTECTED] I'm finding very difficult to work with memory allocation on this platform... :-( I'm trying to reallocate a memory chunk, and my code looks like this: pDBName = (char**)MemPtrNew( sizeof( char*)*2); // i'm allocating memory for only 2 fields, Char *name; // then... in a while rutine: name = (char*)MemPtrNew(appInfo.nameBufLen); StrCopy(name,appInfo.nameP); pDBName[*row]=name; index++; // this keeps the curent no of fields allocated if (index 2) MemPtrResize ( pDBName, sizeof(char*)*index); // the error that i'm caching here is memory locked, but in the documations I read: Call this routine to resize a locked chunk. What's the error is my code? Regards, Iulia -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/ -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: MemPtrResize
From: Orasanu Iuliana [EMAIL PROTECTED] pDBName = (char**)MemPtrNew( sizeof( char*)*2); // i'm allocating memory for only 2 fields, Char *name; // then... in a while rutine: name = (char*)MemPtrNew(appInfo.nameBufLen); Perhaps this needs to be appInfo.nameBufLen + 1? A buffer overrun might trash the start of the next block on the heap, possibly confusing the OS. StrCopy(name,appInfo.nameP); pDBName[*row]=name; Shouldn't this be pDBName[index]=name; ? (although I don't think that on it's own would produce a problem unless *row wasn't valid). I also wouldn't have thought that you need the 'name' variable, since you could save directly into pDBName[]. Which leads to the question as to whether you need to resize the array before you store the new pointer, although it looks like you're always keeping an empty position at the end of the array. What can help to reduce memory fragmentation with this sort of logic is resizing in units of ten or twenty. You have to store the arraysize and current element count separately, but it avoids the repeated resizing involved with this approach. If you're worried about wasted space you can always finish by resizing the array to the final size. index++; // this keeps the curent no of fields allocated if (index 2) MemPtrResize ( pDBName, sizeof(char*)*index); // the error that i'm caching here is memory locked, but in the documations I read: Call this routine to resize a locked chunk. Yeah, I never understood why a Ptr API call would return a locked error, but my guess was that it was an old doc error, or a side-effect of the low-level code that handles ptrs and chunks. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: default application database
From: Vincent Uy [EMAIL PROTECTED] what is the format of the resource file for pdb? are there any documentation on this? From a recent message on this forum: - Original Message - From: Dr. Vesselin Bontchev [EMAIL PROTECTED] I've read the documentation provided by PalmSource http://www.palmos.com/dev/support/docs/protein_books/File_Formats/Intro.html#970019 http://www.palmos.com/dev/support/docs/protein_books/File_Formats/PDBandPRCFormat.html#972428 With the oft-repeated note that there are a number of tools available for working with these files (par, pdbc, ...). Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: shut down problem
From: Scott Erickson [EMAIL PROTECTED]
functions with no problems. then in appstop i try to free the memory with
MemPtrFree(currentSettings); yet the simulator tells me i have an unfreed
chunk (same as the value of the pointer to the variable), and the device
softresets at this point. how can i free the chunk of mem i am using for
this variable? below is some relavent code from my app...
typedef struct{
Boolean monitor;
UInt32 aUpdate;
Char* lastUpdate;
Char* license;
} appSettings;
appSettings *currentSettings;
(in appstop)
MemPtrFree(currentSettings);
What are lastUpdate and license pointing to? Are these allocated
and released elsewhere or could these be the unfreed chunks?
I also have a habit of setting pointers to null after freeing them.
This helps to identify accesses after release and multiple release
attempts. Certainly the unfreed chunk itself shouldn't produce
a reset so you might be looking for a different problem
completely, with the unfreed chunk just a side-issue.
Chris Tutty
--
For information on using the Palm Developer Forums, or to unsubscribe, please
see http://www.palmos.com/dev/support/forums/
Re: Re[2]: T5 database HotSync file size problem
From: Adrien Regimbald [EMAIL PROTECTED] Unfortunately, there is no way around this for now. Perhaps PalmOne will issue a patch, but we don't have any news that one is coming as of yet. The minimum record size on the T5 is actually 512 bytes, which (snip) physicalIndex = logicalIndex / logicalRecordsPerPhysical; physicalOffset = logicalRecordSize * (logicalIndex % logicalRecordsPerPhysical); Of course, it would be a good idea to put this functionality in a wrapper class of some sort, so that you have a new logical family of database functions to perform all of your physical - logical translation for you. Which is exactly what Palm(One or Source) could do and release open-source if they were serious about helping developers overcome this problem. For many apps I suspect that a simple wrapper for the Dm functions would do the job, although I haven't investigated. Has anyone that's taken this step got an opinion on whether it can be implemented as a wrapper lib without code changes? Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to determine available storage space
From: Robert Moynihan [EMAIL PROTECTED] Miguel Angel Sotomayor Hernandez wrote: relevant. Is there a function, like VFSVolumeSize, but to determine the availbale space in storage ram? I need to let the user know when the device is running out of space and calculate how much data You probably need to use MemHeapFreeBytes(), iterating through the individual heaps and tallying them up. Bob. From memory heap 0 is dynamic and heap 1 is storage, but some quick testing should tell you which is which. Remember to keep a good safety margin because things start getting nasty when storage gets tight - it's worth helping users to keep some spare space if you can. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Strange error when trying to intercept sysNotifyVolumeMountedEvent
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] It's just not designed at all with systems programming in mind. Anti-virus programs, device drivers, other deep stuff - hell, even writing libraries isn't exactly trivial there... Actually one of the things that occurred to me after writing that is that you're possibly missing what seems to be a base assumption to Palm OS development - that OS-level programming (which is what you're doing) will be done by licensees working with PalmSource. Much of this isn't documented simply because there are only a handful of apps that will ever be written at that level and the assumption is that those people approach PalmSource, demonstrate the usefulness of their application to the platform and gain NDA-protected access to internal information. I think you're feeling the pain of being isolated from that process, but your organisation certainly has the credibility to sign into an agreement with PalmSource. Now I've never done this, but I've worked with hardware developers for whom it's commonplace. Has your organisation contacted PalmSource with a view to gaining access to their internal resources? Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Strange error when trying to intercept sysNotifyVolumeMountedEvent
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] (yesterday) good old MS-DOS you just intercepted an interrupt vector and did whatever you wanted when the interrupt arrived - Perhaps your memory is different to mine, but I remember all sorts of frustrating, nasty, undocumented problems with apps that just grabbed an interrupt ... and did what they wanted. It was a support and maintenance jungle - but possibly you remember it differently to me. but even Windows 3.x could do cooperative multi-tasking while running over MS-DOS on a 640 Kb machine with a similar CPU... Sigh. Yeah. With the emphasis on 'cooperative'. Anything that broke the rules even slightly brought the whole house of cards down on it's arse. Or is this my memory failing me again :-). I'm not going to pretend that Palm OS was brilliantly designed, but I think your repeated howls to return to the good old days of DOS aren't taking into account that most of what you're working with is better designed and better documented than DOS ever was. I'm not disagreeing that you could do what you wanted in DOS and you can't in Palm OS, but that's life. I remember a huge mountain of things that just didn't work in DOS and do in Palm OS. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Strange error when trying to intercept sysNotifyVolumeMountedEvent
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] Help? Is this the correct way for an application to launch itself with a custom launch code and to check whether this is how it has been launched? If I recall correctly, some apps handle notifications by setting an alarm that will launch their app normally in a second or so. This gives the OS activity associated with the notification time to settle down. It seems to me that in attempting to immediately transfer control to the same app you are setting yourself up for timing and event-queue issues, although I've never attempted to do anything like this. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: DmOpenDatabase returns ErrCantOpen
From: Luc Le Blanc [EMAIL PROTECTED] A user sent me a PDB he created with my program some time ago but that he can't open anymore (although the database is found by SysCreateDataBaseList.). When I try to load it under POSE, the emulator reports a Can't open error, but the PDB nonetheless appears in the FileZ list view. Deleting it with FileZ and retrying eventually succeeds. Even when the PDB is finally loaded into POSE, it can still fail opening. I traced the problem: DmOpenDatabase returns 0x206 (dmErrCantOpen); my program then offers to delete it, but this fails too (how does FileZ do that?). Ideas? I put that 1.6 Kb PDB on: PDBGo shows a header that looks basicly OK (CRID = AURI, Type = Cave, Name = Fieux 2_Cave) with an AppInfoArea at offset 224 and 18 data records, 17 at 56 bytes and the last at 138 bytes. The category list is just binary data, but then you'd get that if you had an AppInfoArea but didn't storing categories (the PDB spec doesn't provide any way to identify this). Does this sound about right? Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Any ideas when Mac version of Eclipse IDE will be out?
From: [EMAIL PROTECTED] A Mac version of Eclipse is already available from eclipse.org but I expect that you were really asking, When will a Mac version of Palm OS Developer Suite be available? I can't answer that one but I can say that I'm open to input from any Mac developers on how important this is to them. Feel free to email me directly. Actually I suspect what's needed is a head-count. I've seen a few people step up from time to time and complain that the Mac community isn't supported but I've got no idea what the size of that community is. As a manager I'd want to know whether that's a dozen vocal developers or a thousand quiet ones before deciding how much money that group gets. I'm not sure how you'd carry out that head-count. Perhaps the easiest way is for someone to establish a Yahoo group and invite subscriptions. That's an easy way for the Mac users to say This is who we are and we want it now. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Please explain why this solution works...
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] Folks, note that all this still doesn't answer Scott's original question. Even if he's computing the string size incorrectly, why does his program behave differently under the simulator and on the real device? It ought to either fail in both cases or work correctly in both cases. Not if the mis-calculation is used as the basis for a string operation. If, for example, that length was used to do a StrNCopy or MemCopy and produced a non-terminated string the behaviour of code that used that string would depend on what followed it in memory. The memory allocation might return clean memory on the simulator and buffer crap on a device (although I'm not as confident about this with the simulator as I was with the emulator because the emulator was known to produce memory allocation with certain characteristics while the Simulator is, in theory, the same OS code as the device). Still, when it comes to how an app functions when the wrong length is given for a buffer it's easy for the answer to depend on trivia such as whether the device was reset before the test, whether the ROM used in the simulator is *exactly* the same version as the code on the device, whether the protocol used is exactly the same (this is, after all, a hangup that follows giving the OS the wrong length for the data being sent - protocol settings might determine whether the connection survives this or gets into an invalid state).. My bet is that it would be possible to identify the source of the difference, but that it might take weeks of work, some seriously low-level debugging and disassemblies of the NetLib code on both the device and the Simulator. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: error checking and debuging on the device
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] It doesnt lock up on the simulator, but it does on the device. This is an unpleasant one... It is not supposed to happen. In practice it will happen for all sorts of things. I've regularly had code work fine on the simulator and lock up the device (generally intermittently). It often seems to be related to the hardware altering the timing of sequences or locking resources differently (the most recent I had was an API call that was sometimes causing a reset during startup on a T3, but was theoretically sound and worked fine in testing. I suspect that that was related to the app attempting to call the API early in the startup sequence while the OS still had the associated resources locked, since when I altered the logic to delay the API call the problem went away). What I found useful over time was to develop a set of logging functions that wrote to a database. This database could then be picked up from the backup folder and analysed. It's slower than debugging, but then I used to spend a lot of time with external devices connected to the serial port and that pretty much trashes your chances of device-based debugging. It's also impossible to trace timing-dependent problems by stepping your code. And finally, you can send a log-enabled copy of your app to that one person who can reproduce that last intermittent crash and have them return the log pdb to you. Try doing *that* over a serial port :-) The logging functions don't have to be complicated, just open a database and just write data strings to it. It's useful to write comma- -separated strings that start with an incrementing value and a log entry type because you can then dump the data out to a text file and import into a spreadsheet or database for analysis. This is useful when you've got twenty thousand lines of debug log and you want to pick out one type of log entry. Including the current tick count is also sometimes useful. I keep meaning to package my log routines as a resource for developers, but it's only an hour or so's work to write your own so... Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: the specified index is out of range.... when using DmReleaseRecord
From: Mishael Kim [EMAIL PROTECTED] //authenticatedH is created and set to false recIndex = 2; authenticatedH = DmNewRecord(gDB, recIndex, recSize); if (authenticatedH) (snip) DmReleaseRecord(gDB, recIndex, true); //authenticatedH is set to true later in code recIndex = 2; recP = MemHandleLock(authenticatedH); DmSet(recP,0,1,true); MemHandleUnlock(authenticatedH); DmReleaseRecord(gDB, recIndex, true); The first creating and initializing works fine...but when i release it the second time i set it to true, i get an error saying that the specified index is out of range. From the API documentation for DmReleaseRecord: Call this routine when you finish modifying or reading a record that youve called DmGetRecord on or created using DmNewRecord. So you're creating a new record, writing some data to it and releasing it - fine. You then tell the OS ok, I'm releasing this again and the OS replies Huh? you've already released that record. Once you've released a record you should stop using any resources associated with that record. If you want to write more data you have to use DmGetRecord to ask for the handle again. Just because you've still got a handle lying around from the last Dm call doesn't mean that it's valid to use it. You've got to remember that while that memory handle might still be available and the lock and write might seem to work that this doesn't necessarily mean that you're following a valid sequence of API calls. What you're doing (New, modify, release, modify, release) would produce bad results in most API's I've seen. What you should be doing is (New, modify, release, Edit, modify, release). What's confusing is that the MemHandleLock looks like it's locking the record for editing, but it's not. Remember that the MemHandleLock isn't a Dm function - calling it doesn't interact with the database API at all. The memory manager and database manager are two layers so you'll have two levels of lock and unlock for each access (I like to wrap these up into one function so that it's impossible to access the handle after releasing the record, but that's just me). Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: one simple question about Delay funcation in Palm OS
From: Alan J. McFarlane [EMAIL PROTECTED] Of course in a application one would just sit at EvtGetEvent when waiting for an IrDA callback, but doing that in a shared library would be bad, no? -- Although I haven't followed this in detail, if you're looking for a way for a shared library to wait in background for an external event one way to do this is to expose a function that the parent app calls as part of their main event loop (with EvtGetEvent set up for the event loop to cycle several times a second as required). This isn't perfect, and is subject to all sorts of uncertainty in that you're depending on the parent app to set up their event loop properly, but it serves the purpose of giving your library background time-slices. I've done this for background serial processing for a shared library used by a family of applications, but I wrote all of the apps that used the library so I had good control over both sides of the relationship. Chris Tutty. -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Flashing Palm Logo
From: Robert Moynihan [EMAIL PROTECTED] As soon as your app starts to process the reset launch code, as the 2nd thing you do, set a flag in memory (I've used the prefs database before, setting a small flag that says I'm handling a reset). When you've finished handling the reset launch, and the last thing you do before returning, clear the flag. The 1st thing you do when processing the reset launch code is to check that flag. If that flag is already set, then it means that you did NOT complete the last reset properly, so it's likely a crash happened, so DON'T do it again. Just bail out and return control to the user. Now that's the kind of excellent technique that's obvious once someone's suggested it. I'll just steal that idea and tuck it away in my things to add to my default startup project pile. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: external databases
From: sachin jesukaran [EMAIL PROTECTED] or if can i convert this *.mdb file to *.pdb file and access it if yes how do i convert a *.mdb file to a *.pdb file. A google search for convert mdb pdb palm gives a long list of hits, the first page of which has some items that look useful. I assume you've checked these out and they're not suitable. Better still, there's searching this archive using http://news.palmos.com/read/search. Do you have some specific questions about the tools you've found in your research or are your requirements so unusual that the tools you've found won't work? Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: reading stored memory
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] Instead, I suggest that on the form that does the scanning, you listen to the nilEvent and scan only a single database when this event occurs. This way you'll give up time to the OS do to other things as well (e.g., alarms to trigger, clocks to be updated, etc.). (snip) If you don't realize how to implement what I described above, ask me by private e-mail; it's easy but a bit long to type here. (I'm using a rather awkward Web interface to access this forum.) There are a number of discussions in the archive relating to background processing that describe how to organise your event loop and avoid hogging the device, many relating to serial comms handling, but providing generic solutions. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: C++ Save to Database, from field (codewarrior) problems
From: Andre Augusto [EMAIL PROTECTED] I have a function to get text from a field and another function to save the text to a database. But when I call the InsertRecord() function, the simulator craches (I also try the code in PDA, with same results.. ). I can't guess what's the problem can anyone help? If you're trying to catch a crash start by checking for null pointers. The InsertRecord() function uses a number of pointers without checking them - the GetField result that's passed to the stratoi, the recP result from the memhandlelock, the db value itself. I'm also not sure about setting index to DmNumRecords to add a record at the end. Although it's probably fine, is there any reason you can't use dmMaxRecordIndex as the documentation for DmNewRecord suggests? Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: DmFindRecordByID API is Failing?? plz help
From: [EMAIL PROTECTED] Its getting proper value in typeID It Display the Alert Message as follows Database:Unique ID not found(Dm 0218) I'm not sure I guess when the new PDB file is installed does its recordID changes?? May be because of that. Is It? This topic has been covered in the archives. From memory the uniqueid generation can be re-initialised by a hard-reset and there are issues associated with restored databases. You might find that the uniqueid isn't suitable for the purpose you are using it, but the archive (found by following the auto-attached link to the palmos dev forums page) has more detail on this. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: memory
From: Orasanu Iuliana [EMAIL PROTECTED]
i tryed to use what u told me.. but still when i debug to see what my
structer contains i can see only invalid pointer!!
i'm using
AppInfoType* foo = (AppInfoType *)MemPtrNew(sizeof(AppInfoType));
MemSet(foo,sizeof(AppInfoType),0);
Heeelllppp!
Do you mean before or after the 'foo =' line? It will
almost certainly be invalid before that. If it's still invalid
before the MemSet line (and since any memory allocation
can fail that should be *foo = ; if (foo) {MemSet} else
{error}) then perhaps your AppInfoType is larger than
64K, or the allocation is failing for some other reason.
Chris Tutty
--
For information on using the Palm Developer Forums, or to unsubscribe, please
see http://www.palmos.com/dev/support/forums/
Re: pdb
From: Orasanu Iuliana [EMAIL PROTECTED] i have worked before with Appforge compiler in the VB programming language, and there was a convertor for .mdb files into .pdb files so.. i have the ideea that the .pbd file have the same structure as the .mdb files.. Nope. Just because one file format can be converted to another says nothing about the similarity of their structure. Pdbs are actually almost completely free-form (and are really just indexed heaps). The converter will have created a table and record structure of it's own design within the pdb framework. but now i'm stuck in this problem: how can i create in CW a new data base... In general you can create a C structure to store your records and then copy this to and from the database record being edited or created. For more information you should read the relevant sections in the Palm OS Companion and Palm OS Reference and come back to this group with specific questions. Reading the source for a sample such as MemoPad will also provide insight as to how the database functions work in practice. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: PDB file format
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] I want to construct a PDB database on a PC, so I need to understand the file format of the PDB files. At the risk of sounding negative, I disagree. I'd prefer to see I want to construct a PDB database on a PC, so I need to know which of the many tools I should use. My requirements are Many have been down this path and released source that does this work. While this sort of low-levl work can be entertaining as an inteelectual exercise, it's really not that productive. As Ben said, there's par, there's also a COM tool called pdbc (yes?) and a commercial COM tool called PDBGo, as well as about a dozen other products I'm not familiar with. They've all been built by people who worked out these byte alignment issues some time ago. OK, I understand that the PDB file has a 80-byte header: (snip) What is supposed to follow is numRecords records, each having the following structure: (snip) Now, if I examine a real PDB file with a hex editor, the stuff after the 80-byte header looks like this: 02 D0 40 F3 20 01 00 00 02 F1 40 F3 20 02 00 00 03 04 40 F3 20 03 00 00 03 16 40 F3 20 04 00 00 In fact, in the particular file I'm examining, the first part (02 D0) *can* be interpreted like that - indeed the first data record resides at offset 0x02D0 from the beginning of the file. But where does 40 F3 come from and what is its meaning? Shrug. If the first record starts at 0x02D0 then my bet is that there's 00 00 before that 02 D0 and that the 40 F3 isn't part of the offset but is the attributes / id pair (40 = dirty? suggesting that this pdb is taken from a backup). But I might be wrong, since getting abyte or two off when deconstructing a binary format can seriously mess up your interpretation. Although I suspect from your bio that I don't need to explain this to you. :-) Don't expect anything of value in the UniqueID field, it's only significant on the device and can be zeros for created PDBs. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: OS 5.x and the sysNotifyAppLaunchingEvent event
From: Dr. Vesselin Bontchev [EMAIL PROTECTED] OK, I have implemented interception of application launching in PalmOS 5.x only, using the Notification Manager and registering to receive notifications for the sysNotifyAppLaunchingEvent. My question is - is there an easy way of determining whether my application has received this notification because some application is being launched - or because some application is quitting? I'm interested only in the launching events; not in the quitting ones. I could, I suppose, check whether the name of the launched application is Launcher - but is there a more intelligent way to do it? I tested for CRID == 'lnch', which is slightly better, but not really an answer. I think the problem is that, as you've realised, you're not being notified of the app's quitting, you're being notified of the launcher app starting up, which is exactly what that notification is meant to tell you. Of course most apps aren't interested in the launcher launch, because it almost always follows an application exit but Palm OS has to tell you so you have to check for it and ignore it. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: OS 5.x and the sysNotifyAppLaunchingEvent event
From: Henk Jonas [EMAIL PROTECTED] Dr. Vesselin Bontchev wrote: I could, I suppose, check whether the name of the launched application is Launcher - but is there a more intelligent way to do it? My opinion: If the CRID is the same who is registered for vchrLauncher(?) or the default launcher, then it's the launcher who gets called. Now that's much better than my answer, I wasn't detecting custom launchers at all (scurries off to change his code). Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: No form event handler error at notification
From: Tam Hanna [EMAIL PROTECTED] I already tried out to modify the code, and it now looks like this: FormPtr myform; myform=FrmInitForm(SyncForm); FrmDrawForm(myform); FrmSetActiveForm(myform); FrmSetEventHandler(myform, HsFormHandleEvent); Why have you chosen this sequence? I ask because I would have thought that FrmSetEventHandler() needed to be set before any events could possibly arise. The documentation for FrmSetActiveForm includes the following: In Palm OS releases earlier than 3.5, this function generated a winEnterEvent for the new form immediately following the winExitEvent for the old form. Starting in Palm OS 3.5, FrmSetActiveForm does not generate the winEnterEvent. The winEnterEvent does not occur until the newly active form is drawn. This suggests that either FrmDrawForm() or FrmSetActiveForm() are capable of raising events and, in the sequence above, this will happen before the event handler for the form is in place. Or am I missing something? Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Dates prior to 1904
From: Clive Walden [EMAIL PROTECTED] Don't get me wrong. I can do the math if I have to. Just surprised I have to roll my own for people still living today. How many applications actually want to do date math on people back to their DoB? While many apps will record different significant dates (birth, graduation, marriage, etc) these are just recorded as fixed points in time. It's rare that an application will need to calculate, for example, how many days a war veteran has lived. After all, if we're talking about missing functions I would have thought that not providing a function to parse floats in text was more of a deficiency than no multi-century date math. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Dates prior to 1904
From: Clive Walden [EMAIL PROTECTED] Chris Tutty wrote: After all, if we're talking about missing functions I would have thought that not providing a function to parse floats in text was more of a deficiency than no multi-century date math. OKs I had no intention of starting one of those software religious war flames. Oops, there's that 'email has no body language' problem again, I should have added a smiley, or a wink, or something to indicate that I was just tossing loose opinion around. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to intercept application launching?
From: Vesselin Bontchev [EMAIL PROTECTED] From: Chris Tutty The real problem is that HackMaster was created because the trap redirection can cause the device to become unstable if multiple apps try to hook into the same trap - I can't OK, I have researched the issue and here is what I've found. (Snip of excellent description of the hack removal problem). Install first... First-Original Install second... Second-First-Original Remove first... Second-[never-never-land]-Original Yup, that's it exactly. Third-party hack managers aside, I see only three possible solutions: I'm not sure how you detect that another app has patched a trap - do you intend to search the memory of all apps for a stored value that equals the address of your own routine? However, if you can do this reliably it's a step forwards. 3) By intercepting SysSetTrapAddress, the application can detect if some other application is trying to intercept one of the system traps it has intercepted. When this happens, it can disable itself by restoring the original system trap address, effectively removing itself from the chain. I like this because it provides good information to the user. My only concern would be that hackmaster programs are likely to hook the same trap so you'll still have to be able to reliably detect and manage multiple patches on the same trap. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to intercept application launching?
- Original Message - From: Vesselin Bontchev [EMAIL PROTECTED] To: Palm Developer Forum palm-dev-forum@news.palmos.com Sent: Thursday, March 24, 2005 4:03 AM Subject: Re: How to intercept application launching? The real problem is that HackMaster was created because the trap redirection can cause the device to become unstable if multiple apps try to hook into the same trap - I can't remember if the instability occurs immediately or if the problem is related to removing your hack. Anyway. If you're going to do this it's very important that you also implement logic to search for third party hack managers and refuse to run if one is installed (because you're messing with the same infrastructure they are). You will also need to give a loud warning to the user that your hack cannot be used in conjunction with any other hack. Annoying, but it's the only option - one hack per device or only use HackMaster compatible apps. OK, I have researched the issue and here is what I've found. Intercepting system traps in PalmOS is pretty much like intercepting interrupts with a TSR program in MS-DOS (yes, I am old enough to have done that). The problem occurs when your program is uninstalled after a program (installed after yours) has intercepted the same thing. Here is what happens. Normally, when you call a system trap, the call goes to the original code in PalmOS, thought a table containing the addresses of the various traps. When your application, let's call it First, intercepts that system trap and an application calls it, the call goes like this: First-Original That's because at installation time your application obtained the address of the original call, saved it somewhere and replaced it with its own address. This way when an application calls this system trap, your program receives control. It does its thing and chains to the original system trap by calling the address it has saved. Now, suppose that a second application (imaginatively named Second) is installed and that it intercepts the same system trap using the same method. Then, when an application calls this system trap, the call goes like this: Second-First-Original So far, so good. If, at this point, the user uninstalls the Second application, everything will be fine (provided that the application remembers to restore the address of the system trap it has intercepted). The problem will occur if the user removes the *First* application while the Second one is still present and active. The First application knows where the original address of the system trap is, and it can even determine that somebody after it has intercepted the same system trap (by intercepting SysSetTrapAddress) - but it has no way of knowing where that somebody has saved the address of the system trap at installation time. So, if at this point the Frist application is removed, we get Second-[never-never-land]-Original In other words, the chain breaks. As soon as some application calls the intercepted system trap, the Second application will receive control. It will do its thing and chain to what it thinks was the original address - except that now it points to a place that doesn't contain the program it used to contain. Now, in MS-DOS, TSR programs usually did not try to uninstall themselves. Even those that did, usually left a small stub in memory that did the interrupt vector chaining. Unfortunately, that's not an option under PalmOS - there, if you delete an application with the memory manager, it is gone; all of it. Third-party hack managers aside, I see only three possible solutions: 1) An application can refuse to uninstall itself, if it detects that another application has intercepted the same system traps it has intercepted - that's relatively easy to implement. However, this will frustrate the user a lot, given that s/he probably won't know what these other applications are, in order to remove them. 2) When uninstalling itself, an application could disable all other applications that have intercepted the same system trap after it. Unfortunately, it has no way of knowing where exactly they remember that they are intercepting something, so they will still think (i.e., tell the user) that they are active - but will, in fact, be disabled. 3) By intercepting SysSetTrapAddress, the application can detect if some other application is trying to intercept one of the system traps it has intercepted. When this happens, it can disable itself by restoring the original system trap address, effectively removing itself from the chain. Neither of these solutions is good enough, but at least they avoid the crashing problems that would otherwise occur. Regards, Vesselin -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/ -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to intercept application launching?
(My apologies for the double post - OE obviously has problems with multiple messages open for posting). From: Vesselin Bontchev [EMAIL PROTECTED] From: Chris Tutty From memory you create a code segment, attach it to your app as a resource (Ben's answer to my question on how to do this should be in the archive), load and lock it and then redirect the trap through that code segment. Ugh. Why is that necessary? Can't you just use SysSetTrapAddress to intercept trap 0xA0A7 (sysTrapsysUIAppSwitch) with a function in your normal code segment? Hmmm... I just found an application that pretty much does what I want (except that it intercepts only application launching and I want to intercept other things too) - and, from looking at the disassembly, it doesn't seem to be doing such stuff. What's it called? It might be the app I wrote :-) Alternatively it's a competing app. They both, of course, use the same underlying mechanisms. Yup, it checks the OS version and registers a notification under PalmOS 5.x. Under the lower versions it just uses SysSetTrapAddress to point trap 0xA0A7 to a subroutine in its code segment. It also uses FtrSet to remember that it is intercepting something (because the interception can be turned off). But it doesn't have additional code segments or anything like that... In order to make sure that its code segment doesn't disappear, it locates it (CurAppDatabase, DmOpenDatabase, DmGetResource) and locks it (MemHandleLock) and protects the database where it resides (DmDatabaseProtect). Sounds more like a different app. I didn't want to keep the whole app in memory so I built a code segment that just does the background stuff and left that behind when the rest of the app unloaded. Probably doesn't matter in the modern devices, but people used to get serious about a few bytes wasted in the old devices. If you were willing to release a code sample that implements the hackmaster-independent mechanism as open source I'd be happy to contribute some of my existing code to it. I just don't have the time to turn that code into a sample and document it. Unfortunately, the nature of the application I'm developing is such that it simply cannot be released as open source (snip) Furthermore, I'm a bit opposed to the concept of the GNU license myself. I mean, if I release something like that, it would mean that anyone who uses any part of it in their application, has to release that application as open source too - and I find such a restriction unacceptable. I agree, as licenses go that sort of virus just forces people to reinvent the same mechanism. Not to mention that Ive seen more confusion about what the the GPL does and does not allow than any other license. I've also seen directly conflicting answers to the same question. If you're going to give something away then you give it away - the coercion associated with the GPL has just confused it's use to the point where it becomes a hindrance rather than an assistance. I've worked in projects where the managers issued a blanket ban on using any tool with a GPL license in it, in spite of the fact that we weren't in violation of the license, simply out of FUD. However, I am perfectly willing to develop a sample demonsrating how to intercept things (i.e., not the actual application I'm writing) and release its source in the public domain (i.e., anyone is allowed to use it in any way they want). Now THAT's what I think Palm OS needs more of. There are a set of quite complex subjects (accessing the addressbook is another) that aren't well documented and for which the original PIM app source code is just confusing. I've been trying to put together some sample source code for a while now but just don't have the time. I'll mail you off-list on this, as I need to investigate the extent to which I'm limited by an NDA on my original Palm OS 5 work in regards to my giving away my Palm OS 4 work. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: DmOpenRef error
From: ULStudent:Robert.Conlon [EMAIL PROTECTED] I no longer get the DmOpenRef error but the second database with DmOpenRef declaration db2 still wont create. We need more detail on the problem to coment 'wont create' is just too vague. Do you get an error? if so, what error? If you remove the first create does the second start working? Have you checked that the DBName is different? How do you know that it doesn't exist? Databases and prcs with the same CRID are often reported as a lump in the Palm OS info screens so you need to use an app such as Filez if you want more info. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: App launch and SysAppLaunch() function.
From: Fred Zheng [EMAIL PROTECTED] Q2. SysUIAppSwitch() works well. But sometimes I need to have the control return to the calling application when finished. So I tries SysAppLaunch(). This is not that pleasant. it always collapses. You are better to stick with SysUIAppSwitch() and return to the calling app by sending a signal of some sort when you launch it. The calling app would identify that it has to return control and would use SysUIAppSwitch() when it's finished. There are a variety of ways to send the signal - you could set a feature or set a value in a shared preference block - but what worked well for me in a similar situation was to define custom launch codes. You then use one of these instead of sysAppLaunchCmdNormalLaunch when calling SysUIAppSwitch(). Your app uses the same logic as for a normal launch, but takes note of anything special it should do when it's finished (one launch code might be appLaunchReturnToApp1, for instance). Alternatively you can use one custom launch code and pass a command block to the app with the special info in it. The programmer's companion has several sections that discuss this and SysUIAppSwitch() also includes some notes. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to intercept application launching?
From: Vesselin Bontchev [EMAIL PROTECTED] And, of course, I'll have to figure out what to do under PalmOS lower than 5.0. I guess, I'll have to write a hack (snip) I don't want my application to depend on the presense of a third-party hack manager... Did that too. In fact, since the client decided that they didn't want Palm OS 4 support I'd be quite happy to sell you the implementation pre-written :-). I get the impression you're writing this yourself and so that probably isn't an option but if it is let me know. You're correct about the hack - the basic theory is documented (partly) and from memory I worked from a test app (DemoHack) which uses HackMaster support. This provided a start while I worked out how to implement the hack without HackMaster. From memory you create a code segment, attach it to your app as a resource (Ben's answer to my question on how to do this should be in the archive), load and lock it and then redirect the trap through that code segment. The real problem is that HackMaster was created because the trap redirection can cause the device to become unstable if multiple apps try to hook into the same trap - I can't remember if the instability occurs immediately or if the problem is related to removing your hack. Anyway. If you're going to do this it's very important that you also implement logic to search for third party hack managers and refuse to run if one is installed (because you're messing with the same infrastructure they are). You will also need to give a loud warning to the user that your hack cannot be used in conjunction with any other hack. Annoying, but it's the only option - one hack per device or only use HackMaster compatible apps. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: How to intercept application launching?
From: Vesselin Bontchev [EMAIL PROTECTED] In fact, since the client decided that they didn't want Palm OS 4 support I'd be quite happy to sell you the implementation pre-written :-). Thanks for the offer but I am writing it myself and I'd like to support as lower an OS version as possible. I think below 5.0 the same implementation should work down to 3.5, possibly 3.0. The trap mechanism is a fairly fundamental part of pre-OS5 Palm OS, as far as I'm aware so it's more of a testing issue than anything else. I also want to intercept the fact that an application has just been beamed to the device, that a HotSync operation has completed, ... For all of these it's a matter of identifying an OS function that will interact at the appropriate point in the process and then hook your code into that sequence. From memory you create a code segment, attach it to your app as a resource (Ben's answer to my question on how to do this should be in the archive), load and lock it and then redirect the trap through that code segment. Ugh. Why is that necessary? Can't you just use SysSetTrapAddress to intercept trap 0xA0A7 (sysTrapsysUIAppSwitch) with a function in your normal code segment? Because (working from what I remember) your normal code segment will be unloaded when your app exits. You need a chunk of code to sit around disconnected from the normal Palm OS app startup and shutdown. It's not as big a deal as it sounds, certainly not much more complicated than the step from single-segment to multi-segment apps. The only reason it took me a couple of days of frustration is that I'd assumed I knew how the linking process worked and couldn't build the app the way I expected. Once I switched my brain off and followed Ben's instructions step by step it just worked. :-) The real problem is that HackMaster was created because the trap redirection can cause the device to become unstable if multiple apps try to hook into the same trap - I can't remember if the instability occurs immediately or if the problem is related to removing your hack. Oh, I don't mind if my application is HackMaster-compatible. What I want to avoid, though, is making it HackMaster-dependent. It's possible, but a bit more work. You have to implement the connection points and resource id rules that the hack management (sorry, System Extension management) apps follow, but you also have to implement the low-level stuff yourself the hard way. You then have to warn the user that using your hack in conjunction with any other hack without a hack manager will cause unpredictable crashes. It's certainly an approach that I think is reasonable - I've had clients ask for the same non-dependency and I'd like to see it better documented. If you were willing to release a code sample that implements the hackmaster-independent mechanism as open source I'd be happy to contribute some of my existing code to it. I just don't have the time to turn that code into a sample and document it. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Very Urgent (Please respond by the end of today, my job depends on it :)
From: Jamie McMullin [EMAIL PROTECTED] 2) The database I was given to do this project with IS packed. Although I'm not sure the format or how to unpack it. It would seem that /30 would delimete each record. The method I've seen is: char *s = PackedRecord-key; UnpackedRecord-prodno = s; s += StrLen(s) + 1; Why this makes no sense at all is that PackedRecord-Key equals something like: '2938012/30AUTO_ADD/30100.20/30' etc. So I don't see how the above method of unpacking will turn that long concatenated string into a bunch of indivdual, accurately un-delimeted records. It won't. That logic is intended for records that are packed with each string null-terminated, ie terminated by a character zero. You can then use StrLen to find the number of characters up to the next null-terminator. You can still use this logic, but you would have to write a function to count the number of characters from the pointer to the next /30 and then use that instead of StrLen. All that pack really means is that instead of storing strings padded with spaces to a constant length as in '2938012 AUTO_ADD 100.20 ' '2938013SOME OTHER DATA100.20 ' '2938013THREE 100.20 ' they're terminated and the next field immediately follows them, as in '2938012\0AUTO_ADD\0100.20\0' '2938013\0SOME OTHER DATA\0100.20\0' '2938013\0THREE\0100.20\0' etc. The main advantage to packing is that you don't waste space padding strings out to the same length, the disadvantage is that you can't just dump the record into a C struct and access it immediately - you've got to copy the fields item by item into the structure you want to use. Part of your problem would seem to be the way this database was created, particularly since numeric data seems to have been turned into strings and might need to be converted back into numbers before you can calculate with it. On the other hand if you're just displaying these values it doesn't matter. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
Re: Very Urgent (Please respond by the end of today, my job depends on it :)
To be correct, this needs changing from: From: Chris Tutty [EMAIL PROTECTED] All that pack really means is that instead of storing strings padded with spaces to a constant length as in '2938012 AUTO_ADD 100.20 ' '2938013SOME OTHER DATA100.20 ' '2938013THREE 100.20 ' to All that pack really means is that instead of storing strings padded with spaces to a constant length as in '2938012 AUTO_ADD\0 100.20 ' '2938013SOME OTHER DATA\0100.20 ' '2938013THREE\0 100.20 ' so that the strings are properly null-terminated when copied into a C struct. And they probably wouldn't be space-padded - might be null-padding, might be whatever crap followed the null terminator in the original edit buffer. Chris Tutty -- For information on using the Palm Developer Forums, or to unsubscribe, please see http://www.palmos.com/dev/support/forums/
