2008-10-22 17:17 UTC+0300 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
+ contrib/rddads/adsx.c
+ added implementation of local indexing support for ADS driver.
Local indexing is used in case server can not do index
creation, for example: index key depends on relation child
Hi,
Szakáts Viktor wrote:
Right now these are the MSVC warnings (last two may
be important):
mysqldd.c(429) : warning C4701: potentially uninitialized local variable
'pItemEof' used
Yes, this could be important. Actually, I left comment about it a few
dozens lines above warning place. I ca
2008-10-22 16:00 UTC+0300 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
* contrib/rddsql/hbsqldd.h
* fixed BYTE * val -> BYTE val[1]
Hi,
Viktor, my code:
typedef struct _MIXKEY
{
ULONG rec;
BYTE notnul;
BYTE val[];
} MIXKEY, *PMIXKEY;
2008-10-22 16:00 UTC+0300 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
* contrib/rddsql/hbsqldd.h
* contrib/rddsql/sqlbase.c
* contrib/rddsql/sqlmix.c
* contrib/rddsql/mysqldd.c
* Finished implementation of national sorting support
* converted // comments to
The message is resent because my last message (without attachment) has
already reached mailing list.
Instead of attaching .zip file, you'll find it at:
http://www.dbtopas.lt/hrb/sdd.zip
==
Hi,
Przemyslaw Czerpak wrote:
Before you will started the work the
Hi,
I found some more RDD in my "lab" folder. It's ADSX. This RDD inherites
from ADS RDD and implements client side index, if server could not solve
index expression, for example: indexing using relations, local
functions, or even indexing using codeblocks.
The code is written in 2006, and
Przemyslaw Czerpak wrote:
In the past someone added hack for EVAL/EVERY to PACK DBF RDD message
passing parameters inside item used to collect results so it's still
overloaded by sth. I can add workaround for it but I cannot promise
that it will work because it's local user invention which came t
Lorenzo Fiorini wrote:
Today we have hbodbc, hbmysql, hbpgsql, hbfbird contrib libs and
probably more in the future.
Their pourpose is to connect Harbour to SQL databases but they use
different ( class ) interfaces.
Hi,
I've also had a problem like this. I've implemented some universal
inte
Przemyslaw Czerpak wrote:
currently exception handling is set (hb_vmSetExceptionHandler() is called)
right before evaluation of main procedure. Shouldn't it be initialized
before _APPMAIN processing or before hb_vmDoInitFunctions()?
It is set before _APPMAIN execution but after INIT procedures
Hi,
currently exception handling is set (hb_vmSetExceptionHandler() is
called) right before evaluation of main procedure. Shouldn't it be
initialized before _APPMAIN processing or before hb_vmDoInitFunctions()?
My problem is that I use _APPMAIN to initialize application environment,
includi
Hi,
after moving from some old Harbour version to 1.0.1 in production
environment my app started to GPF on __DBPACK.
I've debugged code a little bit. After evaluation of
SELF_EVALBLOCK( ( AREAP ) pArea, pBlock )
inside hb_dbfPack(), the result of HB_IS_BLOCK( pBlock ) is FALSE,
because hb
Przemyslaw Czerpak wrote:
here are some words about dlmalloc and multi-thread:
Thread-safety: NOT thread-safe unless USE_LOCKS defined
When USE_LOCKS is defined, each public call to malloc, free,
etc is surrounded with either a pthread mutex or a win32
spinlock (depending o
Przemyslaw Czerpak wrote:
Looks that there in no important difference between DL-MM and BCC-MM in
real MT programs in your system. DL has a little bit better performance.
In my test in Linux DL-MM in MT mode was killing the performance
in MT+ test in comparison to default SUSE malloc when in ST m
But you can still use hb_threadStart(@func()) because flags are optional.
If I move them to second parameter then you have to reserve empty
item before parameters. Now execat syntax of hb_threadStart() is:
hb_threadStart( [ , ] <@sStart> | | ;
[, ] ) -> | NIL
Hi,
tha
Przemyslaw Czerpak wrote:
2008-09-21 23:03 UTC+0200 Przemyslaw Czerpak (druzus/at/priv.onet.pl)
...
* harbour/source/vm/thread.c
+ added support for inheriting visible memvars from current
thread when new thread is created. Memvars in child thread
can be shared with parrent or
Szakáts Viktor wrote:
I'd like to ask for opinions on turning on Harbour
-gc3 switch (native C code generation) for core .prg code.
To me it seems that enabling this switch gives a
noticeable performance boost (mainly to tget, tbrowse,
and other UI classes). The cost is a slightly larger
final e
Przemyslaw Czerpak wrote:
The only reason I see for binding stack preload with "no tls" is that stack
preload also uses inlined Windows like function to access tls. But I see it
as to separate features stack: stack preload and tls access method
(compiler native or system API)?
When compiler n
Hi,
Przemyslaw Czerpak wrote:
Perhaps we should use HB_THREAD_STUB or pass pStack parameter to functions.
I've just implement it for estack.c and hvm.c.
If possible please make tests so we can compare how it effects BCC builds.
Sorry for long delay, I was out of office a little, and I'm not
Hi,
View the controversy it creates, I propose to upload the following rdds:
dbf1.c strict compatibility with C/52 (RT and Types) DBFCDX52.LIB
dbf2.c strict compatibility with CL53 (RT, BM and Types)DBFCDX53.LIB
dbf3.c strict compatibility with VFoxPro (RT and Types) DBFCDXVF.LIB
dbf4.c st
Hi,
+ added .prg function hb_threadSelf() -> | NIL
It returns thread ID pointer variable or NIL when thread was
registered without it (main thread and some 3-rd party threads
created by C API)
Can main thread be terminated after child thread? Can child process wait
for
Hi,
1) memvars are local to the thread including PUBLICs. I use a few
memvars to set some global variables/setting, like, comma separator,
help file name, etc. I do not change it later during program execution.
In MT these PUBLIC memvars are visible only in main thread. Do you
suggest to dup
But if you can create inlined version of above
code for BCC (don't ask me how, I stop seriously using BCC in
DOS days and here some tricks will be necessary) then maybe it
will give enough speed improvement for this compiler.
Hi,
it's possible the goal could be reached, but I'm unable to write
Hi,
Przemyslaw Czerpak wrote:
I guess that if you inline this function then speed of MT Harbour-BCC
programs will be greatly improved. If I understand the above it should
look like (meta code which should be adopted to BCC so it can be inlined):
static __inline__ PHB_STACK hb_stack_ptr( voi
IMHO it would useful to add "stubbed" thread functions to 1.0.1 so
that one can start to add MT code to real apps without requiring to
jump in production with 1.1.0 too early.
Hi,
if we'll do it, we'll have current 1.1.0 code in 1.0.1, so 1.0.1 will
become meaningless. You can use 1.1.0, beca
Hi again,
Looks that the most expensive is TLS access and it reduce the
performance in BCC builds - the cost of ABI in which VM pointer
is not passed to functions :-(.
We can do three things:
1. add some tricks to reduce TLS access like HB_THREAD_STUB in
xHarbour in hvm.c but it makes
Hi,
Przemyslaw Czerpak wrote:
Maybe BCC does not inline InterLocked*() functions or
they are not as efficient as they can be.
Interlocked*() functions are WinAPI functions. It cannot be inlined.
I know that they are Windows API functions but
MSDN says that each compiler should try to inline
I've tested a few memory managers (MM) with tests\memtst.prg. All test is
using BCC55 compiler, singe thread application, FM statistics switched off,
2.2GHz Celeron, 256MB RAM.
Three MM were tested: BCC55, Win32 Local*(), DL MM (see:
http://gee.cs.oswego.edu/dl/html/malloc.html).
The results ar
Hi,
I've tested a few memory managers (MM) with tests\memtst.prg. All test
is using BCC55 compiler, singe thread application, FM statistics
switched off, 2.2GHz Celeron, 256MB RAM.
Three MM were tested: BCC55, Win32 Local*(), DL MM (see:
http://gee.cs.oswego.edu/dl/html/malloc.html).
The
Hi,
I'm back online (after 40 days of no email, no mobile phone, few
civilization). Thanks for great job and 1.0! :)
My co-worker passed me a bug sample:
FUNC Main()
LOCAL aI
aI := {}
AADD(aI, HB_Hash("S", 10))
AADD(aI, HB_HClone(aI[1]))
aI[1,"S"] := 1
aI[2,"S"] -= 1
AADD(aI, HB_
Hi,
Przemyslaw Czerpak wrote:
I also cannot revert the
order of variables stored for HB_P_HASHGEN because it will
change the order in which of user expression are evaluated
f.e. { ++i=>"A", ++i=>"B", ++i=>"C" } so I decided to change
it at runtime and added new function hb_hashAddNew() which
wo
Hi,
and I would expect result "2" (the last one) to be printed.
This we can discus. For sure it should give the same results
as HB_HASH() and now they are reverted.
So we should change HB_HASH() function or the order of variables
stored for HB_P_HASHGEN PCODE.
What do you prefer?
I think it
Massimo Belgrano wrote:
Thanks for explain your vision
The question is "CUI-GUI-WWW without source code modification" is
possible or impossible?
I hope that you and any other harbour GURU or guy can give us a reply
Hi,
it depends on what you mean "CUI-GUI-WWW without source code
modificati
I vote CUI-GUI-WWW without source code modification
Massimo, I think there is no need to do any votes like this. Harbour
does not have any parliament or president. We are creating Harbour, not
some "other" guys. Do you vote for your own actions? Do you vote for
time you get up in the morning,
Hello,
no compile warning/error generated and I would expect result "2" (the
last one) to be printed.
PROC main()
LOCAL aI := {"A" => 1, "A" => 2}
? aI["A"] // prints: 1
RETURN
Best regards,
Mindaugas
___
Harbour mailing list
Harbour@harbou
Hi,
Marek Paliwoda wrote:
If your program is using more than one version of the CRT, some care is
needed when passing certain CRT objects (such as file handles, locales
and environment variables) across DLL boundaries. For more information
on the issues involved and how to resolve them, see P
Hi,
And you consider it as a valid solution ?
What if I wanted to write some C extensions,
mainly to deal with so called "worker threads" ?
Or I would like to use 3-rd party dll which
uses malloc() internally ?
Come on. Be serious. You are a programmer.
I'm not sure I've understood the meanin
Hi,
Sorry guys, but it seems to me you both mix Harbour.dll
with general purpose dll libraries used by a system.
So, what difference do you see between Harbour.dll and "general puporse
dll". Why Harbour.dll could not be a general puporse dll? What are
requirements for dll to be a general pu
Hi,
Szakáts Viktor wrote:
If you go to a task properties, you will get a "Threads" tab,
where you can get the stack with the "Stack" button, where you
should see the names (only for system .dlls). I'm thinking of
something like that.
(I've tried it on a virgin XP SP3, which doesn't have any
de
Hi,
They are not compatible due to different internal naming and
calling convention.
Final user may have many different Harbour applications compiled
by MSVC and BCC using harbour.dll and different names allows to
install both DLLs in system directory.
I understand that, but isn't it the poin
Hi,
I wonder if there is a way to get the name
of a function by address in Windows. Like in
procexp.exe for example.
Do you mean address to function name mapping like .map file info or
debug info attached to executable? I do not find any functionality like
this in procexp.exe.
Best regar
Przemyslaw Czerpak wrote:
2008-07-09 11:57 UTC+0200 Przemyslaw Czerpak (druzus/at/priv.onet.pl)
* harbour/contrib/hbmzip/hbmzip.c
* added setting file attribute in DJGPP builds
! fixed some small typos in OS2 builds reported by David
Thank You, I was going to commit the change, but I
2008-07-08 19:03 UTC+0300 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
* contrib/hbmzip/hbmzip.c
+ implemented file date/time get/set for OS2
; TODO: DOS implementation is still missing
; Implementation is not tested!
David Macias wrote:
[E:\harbour807\harbour\tests]myzip.exe
Hi,
Probably you are using ADS RDD where we have:
if( !pArea->fPositioned )
{
/* TODO: it should return SUCCESS for Clipper compatibility */
return FAILURE;
}
And this is Clipper incompatible behavior local to ADS RDD only
and which probably should be fixed by simple modif
Hi,
Randy Portnoff wrote:
In Clipper, if you attempt to update a record when at EOF (ie. by
mistake), it does NOT cause a runtime error (which is bad!).
It's great! My software uses macro. Macro can write to database. To test
macro evaluation and avoid data to written to database during the
Hi,
I would prefer some free solution giving opportunity to keep
documentation inside source files. I've never used such systems, so I
can not do a good proposal, but I think it will be great if source code
and docs will be in the same place. It would be less chance to forget
update/create d
2008-07-03 16:30 UTC+0300 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
* contrib/hbmzip/hbmzip.c
+ implemented file get attributes and file set attributes for DOS
and OS2
; I've not even tried to compile the code. I have no required OS and
compile enviroment
Hi,
Your code could be a very nice addition to Harbour BTW.
I wonder if you've considered donating it to the project :)
My exception handler code is below. I've never proposed it to be
standard Harbour's code, because my knowledge about exception handling
is limited. I know Linux can get mu
Szakáts Viktor wrote:
I'd like to add a small feature which makes it possible
to pass an app defined string to Harbour, which then gets
added to every hb_out.log entries.
Set( _SET_HBOUTLOGHEADER[, ) ->
This is useful to add important information to the
Harbour generated FM and GPF messages,
Szakáts Viktor wrote:
One possible solution is to allow to compare
pointers to zero using operators (p != 0,
p = 0, p == 0, and even p > 0, p < 0). Comparison
with any other values would return .F., also,
maybe NIL should be also allowed in place of 0,
and we may allow such construct too: IF p ;
Szakáts Viktor wrote:
What to do?
If we agree on 1). We have to delete thunk/harbour-RC1, make a real
tag in tags/harbour-1.0.0RC1 from thunk/harbour revision 8597, add a
new tag in tags/harbour-1.0.0RC2 from thunk/harbour HEAD revision
after we agree we are releasing RC2.
Fully agreed.
One
Szakáts Viktor wrote:
It would be also good to hear other thoughts on this,
and get to an agreement whether to do "branching"
(parallel development) or "tagging" (linear).
Hi,
actually, I do not understand at all what are we doing with
tagging/branching. I've tried to discuss this question
I hope Mindaugas will help us out with some of
his local patches for hbole, maybe it solve this
specific problem too. But of course anyone else
is welcome to address these as well.
Hi,
I'm sorry I've not kept my promise to upload hbole until monday. I'm
very busy these days. I'll be almost off
Przemyslaw Czerpak wrote:
Hi All,
I'll be offline for about two weeks.
best regards,
Przemek
Hi,
bad news for Harbour community, but I hope you'll have a good holiday! :)
Best regards,
Mindaugas
___
Harbour mailing list
Harbour@harbour-project.
Javier wrote:
I think that the better name for the function would be Sprintf() but I
believe that it is not my decision.
Szakáts Viktor wrote:
> As for a name I'd suggest LIBC_SPRINTF() or C_SPRINTF(),
> maybe HB_SPRINTF() as alternatives.
>
> Votes, suggestions and opinions are welcome.
Hi,
Hi,
One way, specific to GTWVT, is to implement a timer
with unique id whose sole function is to refresh the window.
WM_TIMER event will not be delivered until GetMessage/DispatchMessage
loop will be reached. So, it does not solves problem at all.
Or maybe something completely different,
Przemyslaw Czerpak wrote:
Hi All,
I would like to return to the problem with stripping parenthesis
in expression optimizer.
...
If you agree for above modification then I'll commit it ASAP.
Hi,
I see no problem with this commit. We should not limit ourselves to be
compatible with Clipper's
2008-06-10 19:10 UTC+0300 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
* harbour/source/vm/hashfunc.c
* extended functionality of HB_HSet(), HB_HDel() and HB_HDelAt()
to return hash value
___
Harbour mailing list
Harbour@harbour
Hi,
my friend said, he expects HB_HSet(), HB_HDel() and HB_HDelAt() to
return hash, just like Clipper's ADel() or AIns() return array.
This allows to use expressions like:
aHash2 := HB_HSet( HB_HDel( aHash, "old"), "new", "value" )
Any objection to add this small extension?
Best regards,
Szakáts Viktor wrote:
I thought you'll know the solutions to these problems :)
Hi,
if I've understood a problem, you are trying to call dynamic C function
(function with no parameters count known for compiler). I'm using this
solution for some 9 years under Win16, Win32. It's not multiplatf
... I think it would very
nice if there would be a possibility to specify the
_ordinals_ of those format strings:
cName := "Bond"; cCity := "Tokyo"
_SPD( "Hi, I'm %#01s from %#02s.", cName, cCity ) -> "Hi, I'm Bond from
Tokyo"
_SPD( "Hi, I'm from %#02s, my name is %#01s.", cName, cCity ) -> "H
Hi,
one more Clipper/Harbour compatibility issue. Harbour does not set
oErrorObject:Args in case of macro compile syntax error.
Sample:
PROC main()
LOCAL bI
ERRORBLOCK({|e| QOUT(VALTYPE(e:Args))})
bI := &("{|| 'hhh}")
RETURN
Clipper:
A
Harbour:
U
Best regards,
Mindaugas
__
P.S. I also remember Przemek some time ago mentioned some issue about
Linux package versioning. I see last release 0.99.3, on SourceForge
download.
It could be, 0.9.9 is less than 0.99.3, and packages will not be updated
to 0.9.9. So, probably this should be 0.99.9 (or 0.99.4-RC1, to allow
0.99
Phil Barnett wrote:
In SVN, tagging is simply a copy operation, so I placed the tag with the
following command:
$ svn copy
https://harbour-project.svn.sourceforge.net/svnroot/harbour-project/trunk/harbour
https://harbour-project.svn.sourceforge.net/svnroot/harbour-project/trunk/harbour-RC1
I agree, but if we think that MT will come soon we should release a
beta4 not an RC.
Hi,
I really do not care, if it will be 1.0 or 0.1 release, but I want to
have a stable release ASAP and SVN open for new features. I'm waiting
for this for half a year, and I do not want to wait one more ye
Hello,
one more sample code:
--
PROC main()
LOCAL cI := "1"
ProcX(@cI)
? cI
RETURN
PROC ProcX(...)
LOCAL aArg := hb_AParams()
aArg[1] := "2"
RETURN
--
Results are:
xHarbour:
2
Harbour:
1
What is expected result for this code? Is there any way to
Edmer wrote:
Could you please check what is wrong with the following code:
*--- start here ---
PROCEDURE Main
try
? "Base try 0"
try
? "Trying 1"
? "Throwing 1"
throw(1)
catch
...
because it throws me the following compile time errors:
TRY5.PRG(2) Error E0020 Inco
Przemyslaw Czerpak wrote:
So if no one will vote against I plan to change hb_vmFrame()
ASAP and leave any other modifications/optimization after 1.0.
Hi,
I guess no one is against. Please, go on. Thank, You.
Best regards,
Mindaugas
___
Harbour mai
Currently it doesn't compile for MSVC, so in this
form, it really isn't up to release standard, so if
not fixed, it'd better be removed. My vote to update it.
Hi,
it will take a little time to fix code style. So, I hope I'll be ready
to upload it in the beginning of next week. But I don't want
Szakáts Viktor wrote:
Hi folks, Mindaugas,
Do you think we still need hbole?
Isn't hbw32 working for you now?
Brgds,
Viktor
Hi,
I'm not using hbw32. I use my own ole implementation (based on very old
ole code). It is more clean, but less ole features supported (arrays,
parameters by refe
I think you misunderstood (or I didn't explain well),
so the concept was that Harbour zlib wrappers went
to _hbrtl_, all minizip functionality (and only that)
went to hbmzip, and we have hbzlib in core, which BTW,
you can choose to not use at all, by simply replacing
it with 'zlib1.lib' or your ow
The core lib's name would naturally be hbzlib
(like hbpcre), but we already have hbzlib in
contrib, which collides with it.
So, I'd suggest to move strict zlib interface
functions (Harbour level ones that is) from
contrib/hbzlib to source/rtl, and rename the
remaining MiniZip stuff in contrib to
Hello,
there is a compile problem with exhanced codeblocks inside BEGIN/END
SEQUENCE.
PROC main()
BEGIN SEQUENCE
EVAL({||
RETURN 1
})
END SEQUENCE
RETURN
Best regards,
Mindaugas
___
Harbour mailing list
Harbo
I'm not a DOS/Windows user and I expect that when I redirect stderr
to file then for clean builds without any warning this file will be
empty just like in all other compilers I'm using. I know that due to
missing stderr redirection in pure DOS command.com DOS compilers were
using many different ha
Two most important (from my point of view) features:
1) you can do drawing to pdf/png/ps/svg files and Windows DC/X-Win
surface using the same interface function. You just do not need to worry
if your output device is screen or pdf, etc.;
2) multiplatform.
Excellent. I assume you already have
If someone by any chance has a working PDF library,
it could be done (and probably by adjusting the Harbour
level interfaces in HBDOC, too). Such a library
would be a useful contrib by itself, though.
Hi,
I'm using cairo (http://cairographics.org). The new stable 1.6 version
was released last
Hi,
If possible then I would like to ask all of you to make tests with
this new class and your code. It will be the best if you have some
original Clipper code not updated to work also with Harbour or xHarbour.
If sth will be wrong then please inform me.
If all will work correctly then I'll rep
2008-03-15 15:15 UTC+0200 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
* harbour/contrib/hbzlib/zip.c
! fixed compression bug
___
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
Looks like bug in minizip code. I'll try to find a problem in spare time.
Hi,
my bug in minizip code. Just add symbol ";" in the end of
contrib/hbzlib/zip.c line 1054. I have no time to commit the fix now.
Best regards,
Mindaugas
___
Harbour mail
I need to zip / unzip some files and I'm testing hbzlib but I have found
a problem under BCC / WinXP (not tested yet in other environments).
If I compress simple files it creates a valid zip file, but if I
compress even one only exe file (about 600kb of size) the zip is
corrupted.
Tested either
(gdb) bt
#0 0x9442dcec in closedir$UNIX2003 ()
#1 0x0004d4a8 in hb_fsFindFirst (pszFileMask=0x312aa0
"/Users/fiorlo/.notexist/notexist", attrmask=0) at ../../hbffind.c:853
Hi,
looks like missing line in rtl/hbffind.c:hb_fsFindClose():
+if( info->dir )
{
closedir
Przemyslaw Czerpak wrote:
I would prefer release ASAP and changes later.
OK but how many changes we want to introduce before next release
and final 1.0 version? I would like to unblock SVN repository as
soon as possible.
Hi,
saying above I've meant exactly the same as you, i.e.: "I would l
Przemyslaw Czerpak wrote:
but I do not know if it's good idea to touch such important function
just before release. Your opinion?
Hi,
I would prefer release ASAP and changes later.
Best regards,
Mindaugas
___
Harbour mailing list
Harbour@harbour-pro
Hello,
is "#define HB_CLS_NOTOBJECT" before "#include "hbclass.ch"" the right
way to avoid definition of "Init" method for the class?
Best regards,
Mindaugas
___
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/ma
> I have an idea - temporarily add a call to function
> void hb_gcGripTest( PHB_ITEM pItem )
> from for example hb_itemClear() and hb_itemCopy().
Sorry, not so good idea...
Best regards,
Mindaugas
___
Harbour mailing list
Harbour@harbour-project.org
h
Przemyslaw Czerpak wrote:
Poiters cannot be indirectly unlocked during hb_itemPutPtrGC because
using the following code:
HB_ITEM item;
item.type = HB_IT_NIL;
hb_itemPutPtrGC( &item, hb_gcAlloc( size, someFunc ) );
will end with an item storing an unlocked pointer that will be
released prematurely
Przemyslaw Czerpak wrote:
And this message is also confusing because it reports error
in FPT file when the problem is in memo block address stored
in DBF file . It contains repl(" ",9)+chr(10) - Clipper decode
it as number without any validation and this as result it
receives some unexpected bloc
Hi,
Przemyslaw Czerpak wrote:
I know that parameter of of Directory() is passed to system function
unmodified. And the question is a little delicate (not sure is the right
english word to express what I mean), that's why I wanted to say it on
mailing list before changing behavior on SVN.
No
Marek Paliwoda wrote:
Yes, I think it's a good idea. We do not have many problem with GC in
3rd party software, because nobody are using collectible pointers yet,
Do you mean 3rd party software available publicly, or 3rd party software
used privately ? In the latter case it's not true, because
Hi,
No. In fact it's minor modification. It will be enough to initialize
referece counter with 0 and add hb_xRefInc() in hb_itemPut*() functions.
But this modifications should be bound with yet another one (see below).
I was thinking about it, but I didn't know that GC pointers can be
locked
Przemyslaw Czerpak wrote:
But if you hardware generate SIGFPE on moth overflow which is
not caught by CRTL then probably similar problems may appear
in different places, f.e.:
proc main()
local i, v
for i:=v:=1 to 10
v*=2
next
? v
Hi,
The same bug exists in Harbour and xHarbour but only Harbour reports it.
But it can be result of any memory corruption, f.e.:
proc main()
BADCFUNC({})
return
#pragma begindump
#include "hbapiitm.h"
HB_FUNC( BADCFUNC )
{
/* damage 1-st param item type */
Saulius Zrelskis wrote:
OLE include files have their suitable #pragma option directives with
restoring _all_ initial settings, so different alignment seems as if
provided by compiler...
Hi,
I also have a question similar to Saulius. Do you know what structures
needs some specific alignment?
Hi,
Saulius Zrelskis wrote:
Thanks for sharing info. I am intensively using BCC and all related info
is very important.
I'm also using BCC, but haven't find anything wrong until now. That's
why I've made assumption it could be my bug.
One small notice, looking at line:
pEntry = (ULONG*)
Saulius Zrelskis wrote:
> /*
> TODO: BCC compiler error???!!!
> I had a situation when without this donothing() call the whole loop
is not
> executed?! Really, strange! Do we have BCC error here? I hadn't
time to do
> CPU level debugging, but an extra function call influences
gene
Hi,
Francesco Saverio Giudice wrote:
sorry to jump in and if I have misunderstood the context, but it's seems
to me that you need something like background tasks that I have already
implemented (and also ported in harbour but not uploaded).
Or i'm in wrong ?
Background tasks does not solve
Hi,
Ryszard Glab wrote:
I can try to implement this using a message to some object, but in >this
case my whole procedure logic, loops, branching will be hidden. I can
not use DO WHILE !EOF()/DBSKIP()/ENDDO, etc., anymore.
In this case you have two choices:
1) wait for multithread using OS
Ryszard Glab wrote:
On 5 Feb 2008 at 17:21, Mindaugas Kavaliauskas wrote:
I've made a try to implement my ideas. Instead of long descriptions I
present it by sample code:
I still fail what you want to achieve however wouldn't it be
simpler to create an array of objects and
Mindaugas Kavaliauskas wrote:
It's compilers job to process language structures such as loops and ifs
to machine code. And the code could be written to be a single thread
code. So, the question (a little bit philosophical) is, why compiler can
not help to write this single thread code?
2008-02-04 16:40 UTC+0200 Mindaugas Kavaliauskas (dbtopas/at/dbtopas.lt)
* harbour/source/vm/fm.c
! fixed hex dump of fm.log. The previous "hex" dump contained
characters: ( ) . , + - / *
___
Harbour mailing list
Harbour@harbour-p
Hi,
In
your shown code, even in 2nd version, oConn:Receive() is executed, only when
state == 1 or so, but is executed. And you said it does not return until
data is received. Eventually the same situation is created when state
becomes 1.
I've also said, that "server_proc() is called from som
601 - 700 of 736 matches
Mail list logo
