RE: GTM on OSX WAS: [Hardhats-members] more M read questions
I do not know about Mumps but c has built-in functions in the Networking libraries that change from big to little and back htons, htonl, ntohs and ntohl. These are host to network and network to host, short and long. These convert if the number is in the incorrect format and leave it alone if it is already in the right format. Thanks Marc Aylesworth C3I Associates AFRL/IFSE Joint Battlespace Infosphere Team 525 Brooks Rd Rome, NY 13441-4505 Tel:315.330.2422 Fax:315.330.7009 Email: [EMAIL PROTECTED] -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of jae kim Sent: Monday, August 22, 2005 11:58 PM To: hardhats-members@lists.sourceforge.net Subject: Re: GTM on OSX WAS: [Hardhats-members] more M read questions yes, i think so. I liked big-endian better because my Fortran code never had to see Intel chip. Later I had to write some apps to convert big-endian data (Irix) to be read in little-endian supercomputer. Just googled this: http://www.intersystems.com/cache/downloads/documentation/cache5docs/PDFS/AC VE_UsingCVEndian.pdf so, it (Cache) can be run on both big- and little-endian machines but the database file itself has to be converted first, if someone wants to read patient A's file from hospital Y's Sparc by hospital Z's Windows XP server. Somebody should write the code so converting to the different endianism (reading files from different endian machine) can be set up as a parameter at the execution time, negating the above tool. Sorry for the nagging... intersystem... Jae. On 8/22/05, Gregory Woodhouse [EMAIL PROTECTED] wrote: I'm fairly certain the Alpha is big endian. In fact, I think pretty much everything except Intel is big endian. === Gregory Woodhouse [EMAIL PROTECTED] It is foolish to answer a question that you do not understand. --G. Polya (How to Solve It) On Aug 22, 2005, at 7:15 PM, jae kim wrote: Is MUMPS + VistA used on any other big endian machine? Just curious. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: GTM on OSX WAS: [Hardhats-members] more M read questions
That's right, and MUMPS has no bitwise operators. Byte swapping would have to be done (more inefficiently) using normal arithmetic operations. --- Aylesworth Marc A Ctr AFRL/IFSE [EMAIL PROTECTED] wrote: I do not know about Mumps but c has built-in functions in the Networking libraries that change from big to little and back htons, htonl, ntohs and ntohl. These are host to network and network to host, short and long. These convert if the number is in the incorrect format and leave it alone if it is already in the right format. Thanks Marc Aylesworth C3I Associates AFRL/IFSE Joint Battlespace Infosphere Team 525 Brooks Rd Rome, NY 13441-4505 Tel:315.330.2422 Fax:315.330.7009 Email: [EMAIL PROTECTED] -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of jae kim Sent: Monday, August 22, 2005 11:58 PM To: hardhats-members@lists.sourceforge.net Subject: Re: GTM on OSX WAS: [Hardhats-members] more M read questions yes, i think so. I liked big-endian better because my Fortran code never had to see Intel chip. Later I had to write some apps to convert big-endian data (Irix) to be read in little-endian supercomputer. Just googled this: http://www.intersystems.com/cache/downloads/documentation/cache5docs/PDFS/AC VE_UsingCVEndian.pdf so, it (Cache) can be run on both big- and little-endian machines but the database file itself has to be converted first, if someone wants to read patient A's file from hospital Y's Sparc by hospital Z's Windows XP server. Somebody should write the code so converting to the different endianism (reading files from different endian machine) can be set up as a parameter at the execution time, negating the above tool. Sorry for the nagging... intersystem... Jae. On 8/22/05, Gregory Woodhouse [EMAIL PROTECTED] wrote: I'm fairly certain the Alpha is big endian. In fact, I think pretty much everything except Intel is big endian. === Gregory Woodhouse [EMAIL PROTECTED] It is foolish to answer a question that you do not understand. --G. Polya (How to Solve It) On Aug 22, 2005, at 7:15 PM, jae kim wrote: Is MUMPS + VistA used on any other big endian machine? Just curious. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
The VA has been running Cache and VistA servers on a 64-bit platform for years. They have been running on the Alpha chip on VMS with no problems with Cache. - Original Message - From: jae kim [EMAIL PROTECTED] To: hardhats-members@lists.sourceforge.net Sent: Tuesday, August 23, 2005 1:42 AM Subject: Re: GTM on OSX WAS: [Hardhats-members] more M read questions This will be the last email, promise. I guess my terribly unnecessary concern is that when VA hospitals with current 32 bit whatever machine ( are they using Windows on Intel now? or IBM w/ PPC?) want to upgrade their hardware to new 64 bit Intel/AMD chip, (very possible scenario in the near future) the old database file should be read without manipulation by the new platform/chip/M/VistA. Good night. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: GTM on OSX WAS: [Hardhats-members] more M read questions
Bhaskar has been quiet because he was on vacation for two weeks, and has been catching up with his e-mail since his return on Monday morning, August 22. Since his return and diligent attempt to get through things that accumulated in his absence, the number of unread e-mail messages in his Inbox has *increased* by 20. My catching up with hardhats will be slow, and if there is something that you think I should reply to but have not, please do point me to it with an offline e-mail. Thank you very much. I would guess that for someone with a solid Computer Science background and who is comfortable around compilers, run time systems, stack frames, etc., porting GT.M for x86 GNU/Linux to Mac would be on the order of a couple hundred person days. The change from Linux to FreeBSD would be trivial, and indeed members of the GT.M community report being able to hack GT.M to run on FreeBSD on x86. Big endian vs. little endian is a non-issue. I have successfully demonstrated VistA on GT.M on IBM pSeries AIX, which is a big endian machine. Incidentally, the comments about the Alpha being big endian are incorrect - x86 (and the AMD x64), VAX, and Alpha/AXP architectures are all little endian. IBM Power (pSeries) and mainframe (zSeries), Sun SPARC, and HP PA-RISC are big endian. Itanium can be either - Linux and OpenVMS use it as a little endian machine and HP-UX uses it as a big endian machine. -- Bhaskar On Mon, 2005-08-22 at 18:58 -0500, David Sommers wrote: I believe the issue was related to compiler specific optimizations in the C implementation of the M compiler. Bhaskar's been quiet lately but we've discussed this on the list before. I was interested because I simply love my MAC. [KSB] ...snip... --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: GTM on OSX WAS: [Hardhats-members] more M read questions
Yes, I said I thought the Alpha was big endian and I was mistaken. Given that it is essentially the successor to the VAX, I should have known better. --- K.S. Bhaskar [EMAIL PROTECTED] wrote: Incidentally, the comments about the Alpha being big endian are incorrect - x86 (and the AMD x64), VAX, and Alpha/AXP architectures are all little endian. IBM Power (pSeries) and mainframe (zSeries), Sun SPARC, and HP PA-RISC are big endian. Itanium can be either - Linux and OpenVMS use it as a little endian machine and HP-UX uses it as a big endian machine. -- Bhaskar === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: GTM on OSX WAS: [Hardhats-members] more M read questions
On Tue, 2005-08-23 at 11:05 -0400, K.S. Bhaskar wrote: Bhaskar has been quiet because he was on vacation for two weeks, and has been catching up with his e-mail since his return on Monday morning, August 22. Reggie said this in 1977 ;) Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
I certainly do not. --- K.S. Bhaskar [EMAIL PROTECTED] wrote: If anyone ever feels that I get too commercial, please feel free to castigate me. Thank you very much. Regards -- Bhaskar P.S. Kevin, did you ever get your IO working? I think that was what started the thread. === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: GTM on OSX WAS: [Hardhats-members] more M read questions
From what I understand the alpha could be set up to be either big or little endian. Thanks Marc Aylesworth C3I Associates AFRL/IFSE Joint Battlespace Infosphere Team 525 Brooks Rd Rome, NY 13441-4505 Tel:315.330.2422 Fax:315.330.7009 Email: [EMAIL PROTECTED] -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Greg Woodhouse Sent: Tuesday, August 23, 2005 11:46 AM To: hardhats-members@lists.sourceforge.net Subject: RE: GTM on OSX WAS: [Hardhats-members] more M read questions Yes, I said I thought the Alpha was big endian and I was mistaken. Given that it is essentially the successor to the VAX, I should have known better. --- K.S. Bhaskar [EMAIL PROTECTED] wrote: Incidentally, the comments about the Alpha being big endian are incorrect - x86 (and the AMD x64), VAX, and Alpha/AXP architectures are all little endian. IBM Power (pSeries) and mainframe (zSeries), Sun SPARC, and HP PA-RISC are big endian. Itanium can be either - Linux and OpenVMS use it as a little endian machine and HP-UX uses it as a big endian machine. -- Bhaskar === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Most MUMPS implementations have no problem with binary data. It's old utilities that are oriented to text-only data that might have a problem with it. GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. We use this capability many thousands of times a day in VMACS and M2Web. Ruben wrote: On Sun, 2005-08-21 at 15:13 -0400, smcphelan wrote: Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. This is all beyond me because all data is just data. ASCI, Binary, all the same thing. No matter how you look at it, a byte can only have one of 256 representations. The rest is all interpretation. Ruben --- Jim Self Systems Architect, Lead Developer VMTH Computer Services, UC Davis (http://www.vmth.ucdavis.edu/us/jaself) --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
. GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. Hmm - Does Cache not have this capability? thurman -Original Message- From: [EMAIL PROTECTED] [mailto:hardhats- [EMAIL PROTECTED] On Behalf Of Jim Self Sent: Monday, August 22, 2005 3:13 AM To: hardhats-members@lists.sourceforge.net Subject: Re: [Hardhats-members] more M read questions Most MUMPS implementations have no problem with binary data. It's old utilities that are oriented to text-only data that might have a problem with it. GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. We use this capability many thousands of times a day in VMACS and M2Web. Ruben wrote: On Sun, 2005-08-21 at 15:13 -0400, smcphelan wrote: Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. This is all beyond me because all data is just data. ASCI, Binary, all the same thing. No matter how you look at it, a byte can only have one of 256 representations. The rest is all interpretation. Ruben --- Jim Self Systems Architect, Lead Developer VMTH Computer Services, UC Davis (http://www.vmth.ucdavis.edu/us/jaself) --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
Off hand, I don't know, but members of this list do seem to have a tendency to plug GT.M (presumably because it is open source). Personally, I think we'd all benefit from a little more vendor neutrality. --- Thurman Pedigo [EMAIL PROTECTED] wrote: . GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. Hmm - Does Cache not have this capability? thurman === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 11:18, Greg Woodhouse wrote: Off hand, I don't know, but members of this list do seem to have a tendency to plug GT.M (presumably because it is open source). Personally, I think we'd all benefit from a little more vendor neutrality. Good Morning! I don't know what vendor neutral is. When something is Free Software, it means that the programmer and user have control of their destiny. When the software is not free, it means they have zero control, and are almost always abused. Should this reality be ignored because of someones private interests or feature set? I can't speak for others, but I always do whats best for me and my clients. And that means I use Free Software always. Ruben --- Thurman Pedigo [EMAIL PROTECTED] wrote: . GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. Hmm - Does Cache not have this capability? thurman === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
What $P change is that? In any case, I agree with Kevin that I/O in MUMPS could be simplified. That being said, this is a stumbling block in any language becaue the user needs both the capability of reading (up to) a fixed number of bytes and scanning the input stream for complete lines of text (along the lines of the library function scanf in C). It's not uncommon to see entire books devoted to the intricacies of I/O in a particular language. --- Nancy Anthracite [EMAIL PROTECTED] wrote: I thought that was part of the purpose of $P change in the GTM code, to get it to recognize the TCP processes and not cut them off when something that would otherwise be recognized as a control character was sent. Can you leverage that somehow? On Sunday 21 August 2005 09:46 am, Kevin Toppenberg wrote: The read command in M seems to be the most complicated function it has. I am trying to perform a binary read. I do it this way: read blockIn#255 The problem is that as I debug the code, $length(blockIn) does not always=255. I think this is because sometimes the stream contains a terminator, such as a #13 etc. How do do a read that ignores the usual terminators? Thanks Kevin --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members -- Nancy Anthracite --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
In the latter case, there is the problem that M programmers have a propensity for using sentinel values like ^ to delimit data items, but the first problem is much more serious. The only sure fire way to get data moved somewhere is with something like fread or fwrite. You must know how much data your moving because any character (or even combination of chars) can be part of a string/binary Ruben === Gregory Woodhouse [EMAIL PROTECTED] The most incomprehensible thing about the world is that it is at all comprehensible. --Albert Einstein (1879-1955) On Aug 21, 2005, at 1:26 PM, Ruben Safir wrote: On Sun, 2005-08-21 at 12:57 -0700, Chris Richardson wrote: Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character In terms of storing and retrieving data, it shouldn't matter. We've been dealing with 7 bit encoded attachments for a decade in email. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
I thought that was part of the purpose of $P change in the GTM code, to get it to recognize the TCP processes and not cut them off when something that would otherwise be recognized as a control character was sent. Can you leverage that somehow? On Sunday 21 August 2005 09:46 am, Kevin Toppenberg wrote: The read command in M seems to be the most complicated function it has. I am trying to perform a binary read. I do it this way: read blockIn#255 The problem is that as I debug the code, $length(blockIn) does not always=255. I think this is because sometimes the stream contains a terminator, such as a #13 etc. How do do a read that ignores the usual terminators? Thanks Kevin --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members -- Nancy Anthracite --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
First of all, this is not a WorldVistA list, this is Hardhats. If WorldVistA chooses to focus entirely on GT.M that is their choice, but not everyone running (or interested in running) VistA will use GT.M, and the intent of this list is to focus on VistA infrastructure regardless of the platform. --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 11:18, Greg Woodhouse wrote: Off hand, I don't know, but members of this list do seem to have a tendency to plug GT.M (presumably because it is open source). Personally, I think we'd all benefit from a little more vendor neutrality. Good Morning! I don't know what vendor neutral is. When something is Free Software, it means that the programmer and user have control of their destiny. When the software is not free, it means they have zero control, and are almost always abused. Should this reality be ignored because of someones private interests or feature set? I can't speak for others, but I always do whats best for me and my clients. And that means I use Free Software always. Ruben --- Thurman Pedigo [EMAIL PROTECTED] wrote: . GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. Hmm - Does Cache not have this capability? thurman === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Right (no pun intended). The problem, of course, is that a # read in MUMPS doesn't necessarily behave like an fread. If it did, life would be much simpler. Or would it? There is still the problem ot text I/O, and unlike most other languages, MUMPS provides no standard mechanism of for linking to runtime libraries. That means that programmers tend to use the basic I/O facilities provided at the language level, and not library functions like scanf (or fread) in C. Fileman and Kernel attempt to address this problem by providing routines like ^DIR (the Fileman reader) and tools like the Kernel device handler, but the solution is not really ideal. My point of view is that programmers should not have to do extra work or rely on libraries or applications over and above the basic language environment to perform basic I/O tasks. In addition, languages need to provide support for binary files, pipes and FIFOs, TCP channels and the like. MUMPS is very (7-bit) text-centric, essentially by design. --- Ruben Safir [EMAIL PROTECTED] wrote: In the latter case, there is the problem that M programmers have a propensity for using sentinel values like ^ to delimit data items, but the first problem is much more serious. The only sure fire way to get data moved somewhere is with something like fread or fwrite. You must know how much data your moving because any character (or even combination of chars) can be part of a string/binary Ruben === Gregory Woodhouse [EMAIL PROTECTED] The most incomprehensible thing about the world is that it is at all comprehensible. --Albert Einstein (1879-1955) On Aug 21, 2005, at 1:26 PM, Ruben Safir wrote: On Sun, 2005-08-21 at 12:57 -0700, Chris Richardson wrote: Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character In terms of storing and retrieving data, it shouldn't matter. We've been dealing with 7 bit encoded attachments for a decade in email. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Guys, Kevin happens to be trying to code this at the moment and he is running GTM on his server. He is actually trying to write something cross-platform ultimately, so you can both pull in your horns. We are looking for a scanning solution that will work for everyone, and Kevin is putting a lot of time and effort in on this as are others working on this, so be nice, please. On Monday 22 August 2005 12:04 pm, Greg Woodhouse wrote: First of all, this is not a WorldVistA list, this is Hardhats. If WorldVistA chooses to focus entirely on GT.M that is their choice, but not everyone running (or interested in running) VistA will use GT.M, and the intent of this list is to focus on VistA infrastructure regardless of the platform. --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 11:18, Greg Woodhouse wrote: Off hand, I don't know, but members of this list do seem to have a tendency to plug GT.M (presumably because it is open source). Personally, I think we'd all benefit from a little more vendor neutrality. Good Morning! I don't know what vendor neutral is. When something is Free Software, it means that the programmer and user have control of their destiny. When the software is not free, it means they have zero control, and are almost always abused. Should this reality be ignored because of someones private interests or feature set? I can't speak for others, but I always do whats best for me and my clients. And that means I use Free Software always. Ruben --- Thurman Pedigo [EMAIL PROTECTED] wrote: . GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. Hmm - Does Cache not have this capability? thurman === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members -- Nancy Anthracite --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 12:04, Greg Woodhouse wrote: First of all, this is not a WorldVistA list, this is Hardhats. If WorldVistA chooses to focus entirely on GT.M that is their choice, but not everyone running (or interested in running) VistA will use GT.M, and the intent of this list is to focus on VistA infrastructure regardless of the platform. I understand this, but what are you going to do when the seemingly majority of developers on the list are making the rational decision to use the Free Software product. Should the letters G T and M be expunged from the list? You can't make VistA work without the right kind of database and the two are strongly intertwined. I wish I could run it with MYSQL, but I can't. It's difficult for me to understand any complaint in regard to the lists members focusing on GT.M which is the Free Software database backend that they are using. It sound to me a little bit like complaining about gentrification. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 12:15, In addition, languages need to provide support for binary files, pipes and FIFOs, TCP channels and the like. MUMPS is very (7-bit) text-centric, essentially by design. Writing an encoding scheme should be fairly straight forward. 7 bit to 8 bit is exactly what most mail programs do. The problem is knowing how much data your drawing in prior to moving the data. And there has to be some kind of table which describes the basic operating environment to determine if it is 7,6,8, or even 10 bit chars (although I'll bet it is 8). Ruben --- Ruben Safir [EMAIL PROTECTED] wrote: In the latter case, there is the problem that M programmers have a propensity for using sentinel values like ^ to delimit data items, but the first problem is much more serious. The only sure fire way to get data moved somewhere is with something like fread or fwrite. You must know how much data your moving because any character (or even combination of chars) can be part of a string/binary Ruben === Gregory Woodhouse [EMAIL PROTECTED] The most incomprehensible thing about the world is that it is at all comprehensible. --Albert Einstein (1879-1955) On Aug 21, 2005, at 1:26 PM, Ruben Safir wrote: On Sun, 2005-08-21 at 12:57 -0700, Chris Richardson wrote: Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character In terms of storing and retrieving data, it shouldn't matter. We've been dealing with 7 bit encoded attachments for a decade in email. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
I'm very familiar with electronic mail. I'll have to check to see if it's still there, but I may be responsible for a paltry single phrase in RFC 2821 (though I'd like to think that my participation in the WG mailing lists amounted to a bit more than that). --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 12:15, In addition, languages need to provide support for binary files, pipes and FIFOs, TCP channels and the like. MUMPS is very (7-bit) text-centric, essentially by design. Writing an encoding scheme should be fairly straight forward. 7 bit to 8 bit is exactly what most mail programs do. The problem is knowing how much data your drawing in prior to moving the data. And there has to be some kind of table which describes the basic operating environment to determine if it is 7,6,8, or even 10 bit chars (although I'll bet it is 8). Ruben --- Ruben Safir [EMAIL PROTECTED] wrote: In the latter case, there is the problem that M programmers have a propensity for using sentinel values like ^ to delimit data items, but the first problem is much more serious. The only sure fire way to get data moved somewhere is with something like fread or fwrite. You must know how much data your moving because any character (or even combination of chars) can be part of a string/binary Ruben === Gregory Woodhouse [EMAIL PROTECTED] The most incomprehensible thing about the world is that it is at all comprehensible. --Albert Einstein (1879-1955) On Aug 21, 2005, at 1:26 PM, Ruben Safir wrote: On Sun, 2005-08-21 at 12:57 -0700, Chris Richardson wrote: Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character In terms of storing and retrieving data, it shouldn't matter. We've been dealing with 7 bit encoded attachments for a decade in email. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile
Re: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 12:41, Greg Woodhouse wrote: I'm very familiar with electronic mail. I'll have to check to see if it's still there, but I may be responsible for a paltry single phrase in RFC 2821 (though I'd like to think that my participation in the WG mailing lists amounted to a bit more than that). Hey - Then we can just lift the code and translate it! Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Right after you tell me about the bitwise operators in MUMPS! Actually, though you can't do bit arithmetic in MUMPS, you do have $A() and $C() so its quite possible to perform ordinary arithmetic on character values, so something like base64 shouldn't be too hard. Fileman even provides a basic hex encoding tool, unfortunately misnamed $$HTML^DILF. It works like this: W $$HTML^DILF(This^is^a^delimited^string,1) This#94;is#94;a#94;delimited#94;string (a second argument of -1 performs the inverse operation). I've used this call a number of times, particularly in input/output transforms. --- Ruben Safir [EMAIL PROTECTED] wrote: Hey - Then we can just lift the code and translate it! Ruben === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Oops. that didn't wor'k. I'm using a web interface and the 94's were translated to ^. I wa wrong on two counts: the encoding is decimal, not hexadecimal, and the encoding used is the standard one for numeric entities in XML/HTML (meaning, BTW, that the name of the call is less of a misnomer than I thought. What I was expecting is the familiar URL encoding where the space (decimal 32, hexadecimal 20) becomes %20 and so forth. --- Greg Woodhouse [EMAIL PROTECTED] wrote: Right after you tell me about the bitwise operators in MUMPS! Actually, though you can't do bit arithmetic in MUMPS, you do have $A() and $C() so its quite possible to perform ordinary arithmetic on character values, so something like base64 shouldn't be too hard. Fileman even provides a basic hex encoding tool, unfortunately misnamed $$HTML^DILF. It works like this: W $$HTML^DILF(This^is^a^delimited^string,1) This^is^a^delimited^string (a second argument of -1 performs the inverse operation). I've used this call a number of times, particularly in input/output transforms. --- Ruben Safir [EMAIL PROTECTED] wrote: Hey - Then we can just lift the code and translate it! Ruben === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
Greg Woodhouse wrote: Off hand, I don't know, but members of this list do seem to have a tendency to plug GT.M (presumably because it is open source). Personally, I think we'd all benefit from a little more vendor neutrality. I am not a vendor and neither is GT.M. I mention GT.M (and Linux and Apache) because that is what I use and know best and because GT.M on Linux is a completely Open Source (Free) implementation of MUMPS that can form the basis for totally free installations of VistA and other MUMPS based information systems and applications, including M2Web and VMACS, At the moment, GT.M is the only Free (Open Source) MUMPS implementation that has been taken seriously enough by VistA developers to make VistA work on it. I believe that MUMPS_V1 implements enough of the MUMPS standard that VistA could be made to run on it quite well, but that has not been done yet as far as I know. --- Thurman Pedigo [EMAIL PROTECTED] wrote: . GT.M, for instance is quite capable of uploading and downloading binary data, such as images, just fine. It can also easily hand off that task to utilities available in the Linux environment. Hmm - Does Cache not have this capability? thurman --- Jim Self Systems Architect, Lead Developer VMTH Computer Services, UC Davis (http://www.vmth.ucdavis.edu/us/jaself) --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
-Original Message- From: [EMAIL PROTECTED] [mailto:hardhats- [EMAIL PROTECTED] On Behalf Of Ruben Safir Sent: Monday, August 22, 2005 9:53 AM To: hardhats-members@lists.sourceforge.net Subject: RE: [Hardhats-members] more M read questions I can't speak for others, but I always do whats best for me and my clients. Ruben Glad to have that clarified/t --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
--- Jim Self [EMAIL PROTECTED] wrote: Greg Woodhouse wrote: Off hand, I don't know, but members of this list do seem to have a tendency to plug GT.M (presumably because it is open source). Personally, I think we'd all benefit from a little more vendor neutrality. I am not a vendor and neither is GT.M. No, but Fidelity is. I mention GT.M (and Linux and Apache) because that is what I use and know best I can't fault you for that. and because GT.M on Linux is a completely Open Source (Free) implementation of MUMPS that can form the basis for totally free installations of VistA and other MUMPS based information systems and applications, including M2Web and VMACS, I'm less impressed by this argument. First of all, totally free is an illusion. You may not spend money on licensing fees, but if it takes you 10 or 100 or 1,000 hours of work to install and configure the system, that is a cost. For some people, paying for a license may be a better option if they can recoup the cost in other ways (such as less time being required to install, configure and maintain the product). Second, this is really an ideological program. Open source may be a good thing (and I think it is), but is it being touted because it's the right way to do things or because it's the cheapest? At the moment, GT.M is the only Free (Open Source) MUMPS implementation that has been taken seriously enough by VistA developers to make VistA work on it. Perhaps so. But reason may only be that it has become something of a juggernaut -- people put their efforts into GT.M because that is where other people are putting their efforts, and they want to see the time and effort they put into it make a difference. Again, this is not necessarily a bad thing. If there is no compelling technical reason to opt for an alternate product, people will likely not do so. This does not, however, imply that GT.M is technically superior (or inferior) to any of its alternatives. In a sense, it is simply the product that won. I believe that MUMPS_V1 implements enough of the MUMPS standard that VistA could be made to run on it quite well, but that has not been done yet as far as I know. And, as I suggested above, there may be no compelling reason to do so. If I am the only developer interested in working with MUMPS_V1 and I have no reason to believe that a critical mass of developers will share a similar interest, I have no real incentive to do so. I've actually thought about inquiring into whether porting MUMPS_V1 to OS X might be an option, but I also have limited time and other things I'd rather do with the time I do have. No doubt we can all say essentially the same thing. === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 14:26 -0700, Greg Woodhouse wrote: I'm less impressed by this argument. First of all, totally free is an illusion. You may not spend money on licensing fees, but if it takes you 10 or 100 or 1,000 hours of work to install and configure the system, that is a cost. That's the wrong totally free. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Normal email is text based. In mime encoding, it handles binary files as attachments by surrounding them with a string of ASCII characters guaranteed not to be included in the content. This is a standard feature of web browsers used to upload HTML forms that can include binary data such as image files. Ruben Safir [EMAIL PROTECTED] wrote: In the latter case, there is the problem that M programmers have a propensity for using sentinel values like ^ to delimit data items, but the first problem is much more serious. The only sure fire way to get data moved somewhere is with something like fread or fwrite. You must know how much data your moving because any character (or even combination of chars) can be part of a string/binary --- Jim Self Systems Architect, Lead Developer VMTH Computer Services, UC Davis (http://www.vmth.ucdavis.edu/us/jaself) --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 14:26 -0700, Greg Woodhouse wrote: Open source may be a good thing (and I think it is), but is it being touted because it's the right way to do things or because it's the cheapest? I don't know what Open Source is exactly, but in the case of Free Software, its the right way to do things. Its also usually the least expensive way as well, but that is just a side affect of it being a Free Software program. The freer things are, in most things, the more money you make and the less you pay for a unit. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 14:26 -0700, Greg Woodhouse wrote: I've actually thought about inquiring into whether porting MUMPS_V1 to OS X might be an option, but I also have limited time and other things I'd rather do with the time I do have. No doubt we can all say essentially the same thing. That shouldn't be too hard. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
Maybe not, but it's still not at the top of my priorities list. --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 14:26 -0700, Greg Woodhouse wrote: I've actually thought about inquiring into whether porting MUMPS_V1 to OS X might be an option, but I also have limited time and other things I'd rather do with the time I do have. No doubt we can all say essentially the same thing. That shouldn't be too hard. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
That's one perspective, but to others it might be a business decision. In the case of VistA, a hypothetical health care provider may opt for an open source solution because it believes that they money saved in licensing fees will not be offset, say, by increased support costs. Now, it may well be that the open source product is also the best product available (Apache comes to mind here), and it may well be that the overall cost of ownership is less for an open source product in these cases. But the point is that, whether you agree or disagree, a reasonable person may opt for an open source solution because it seems to be the best option available based on purely financial considerations. It is also possible that a developer or (less likely, but still possible) an organization may opt for an open source solution based primarily on ethical considerations or matters of public policy. These are all issues worth discussing, but they are also out of scope for this list, so I'll leave it at that. --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 14:26 -0700, Greg Woodhouse wrote: Open source may be a good thing (and I think it is), but is it being touted because it's the right way to do things or because it's the cheapest? I don't know what Open Source is exactly, but in the case of Free Software, its the right way to do things. Its also usually the least expensive way as well, but that is just a side affect of it being a Free Software program. The freer things are, in most things, the more money you make and the less you pay for a unit. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Gregory Woodhouse [EMAIL PROTECTED] My point of view is that programmers should not have to do extra work or rely on libraries or applications over and above the basic language environment to perform basic I/O tasks. In addition, languages need to provide support for binary files, pipes and FIFOs, TCP channels and the like. If you are just using the language and system features available, you need to learn exactly how they work and how they can be used to provide the most effective solutions to the problems you take on. In areas outside of the coverage of standard MUMPS, you need to make allowances for implementation specific variations and features. Going with a lowest common denominator approach, especially in areas involving I/O, generally has such degraded performance as to be simply unacceptable. The VistA kernal developers have done an admirable job of providing a common interface for many essential features, but I think that pretty much stops short of handling binary data. You keep making noises like you want to redesign the MUMPS language. Perhaps you should get involved with the development of GT.M or MUMPS_V1 or Kevin O'Kane's MUMPS, or even better with all three, and bring them up to the level of functionality and quality that you need - the source code is available and Free to be improved upon. With the demise of the MUMPS Users Group (aka MUMPS Technology Association aka MTA) and of the MUMPS Development Committee (MDC) and of most of the vendors that followed Standard MUMPS, especially Datatree, Micronetics, and Digital Equipment Corporation (DEC), the days of nagging the vendors for enhancements to the language are pretty much all gone - the vendors essentially do not exist any more. Now that we have Free implementations of MUMPS (free as in freedom, not free beer), anyone who really knows what they are talking about has the option of fixing the problems directly. MUMPS is very (7-bit) text-centric, essentially by design. No it's not. It is character oriented and for a very long time all standard MUMPS implementations that I know of have supported at least 8-bit characters and some 16-bit. --- Jim Self Systems Architect, Lead Developer VMTH Computer Services, UC Davis (http://www.vmth.ucdavis.edu/us/jaself) --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 18:10, Greg Woodhouse wrote: That's one perspective, but to others it might be a business decision. In the case of VistA, a hypothetical health care provider may opt for an open source solution because it believes that they money saved in licensing fees will not be offset, say, by increased support costs. To be fair Greg, frankly, the people making these decisions aren't even qualified to make such choices. It cost $200K/year from Siemans for a barely usable system in which it take them 6 months to simply change a label on a form. Just to change the F*^4g label so that a user knows what the heck s/he's looking at. So what kind of support are we talking about here? We all make our money on the support. The only difference is, with a Free Software system your not locked into a single vendor. So who is making these decisions? Now, it may well be that the open source I think you mean Free Software ... product is also the best product available (Apache comes to mind here), and it may well be that the overall cost of ownership is less for an open source product in these cases. But the point is that, whether you agree or disagree, a reasonable person may opt for an open source solution because it seems to be the best option available based on purely financial considerations. It is also possible that a developer or (less likely, but still possible) an organization may opt for an open source solution based primarily on ethical considerations or matters of public policy. These are all issues worth discussing, but they are also out of scope for this list, so I'll leave it at that. :) Ever read the book Town Hall Ruben --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 14:26 -0700, Greg Woodhouse wrote: Open source may be a good thing (and I think it is), but is it being touted because it's the right way to do things or because it's the cheapest? I don't know what Open Source is exactly, but in the case of Free Software, its the right way to do things. Its also usually the least expensive way as well, but that is just a side affect of it being a Free Software program. The freer things are, in most things, the more money you make and the less you pay for a unit. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 18:00, Greg Woodhouse wrote: Maybe not, but it's still not at the top of my priorities list. That's COMPLETELY understandable. Ruben --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 14:26 -0700, Greg Woodhouse wrote: I've actually thought about inquiring into whether porting MUMPS_V1 to OS X might be an option, but I also have limited time and other things I'd rather do with the time I do have. No doubt we can all say essentially the same thing. That shouldn't be too hard. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
--- Jim Self [EMAIL PROTECTED] wrote: Gregory Woodhouse [EMAIL PROTECTED] You keep making noises like you want to redesign the MUMPS language. Perhaps you should get involved with the development of GT.M or MUMPS_V1 or Kevin O'Kane's MUMPS, or even better with all three, and bring them up to the level of functionality and quality that you need - the source code is available and Free to be improved upon. There are a number of reasons I'm not especially interested in getting involved with GT.M development right now. I have (seriously) considered it, but I have seen little interest expressed on this list in addressing the types of issues I've attempted to call attention to, and supporting the existing code base seems to be the biggest (even only) concern of most people on this list. Intel may have released a series of chips backward compatible with the 8088, and that strategy worked well for a time, but eventually they had to break with the past and deverlop new lines of chips that are no longer backward compatible with their old 8-bit CPU with its 8-bit bus. I don't think VistA's really any different, but we're kind of stuck at the 80286 stage. No one seems ready to make the leap to the 80386, much less less the Pentium 4. === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
From: "Ruben Safir" Monday, August 22, 2005 6:03 PM I don't know what Open Source is exactly, but in the case of Free Software, its the "right" way to do things. Its also usually the least expensive way as well, but that is just a side affect of it being a Free Software program. The freer things are, in most things, the more money you make and the less you pay for a unit. RubenYou have missed a very critical difference in the concept of Open Source,that is that the source code is available. Free software, the stuff that isgiven away, does not always incorporate the source code. The fact that Ihave access to the source code permits me to know the limitations of theprogram, at the very least (Microsoft model) and to add function to theprogram in most cases.As a result, I do not support the Open Source concept as "Free Software" butrather as software that can grow, change and not become orphaned by theauthor. As has been noted in this and other lists, the cost of support isthe real expense in software. The "least expensive way" is the method thatresults in the longest lifespan of the system and the lowest cost of supportover the life of the product. Please do not confuse this with the cost ofpurchase or implementation.Lastly, what is meant by the statement: "The freer things are, in mostthings, the more money you make and the less you pay for a unit." Who isreferenced by the "you".Mike
Re: [Hardhats-members] more M read questions
You have missed a very critical difference in the concept of Open Source, that is that the source code is available. Free software, the stuff that is given away, does not always incorporate the source code. The fact that I have access to the source code permits me to know the limitations of the program, at the very least (Microsoft model) and to add function to the program in most cases. I believe your confusing Free Software with the marketing speech created by Bruce Perens. It's Free Software, like in Free Market, and it has a specific definition which is detailed here http://www.gnu.org/philosophy/free-sw.html On the other hand, the term Open has been badly mangled by MS, Sun and others. Its very confusing to use the term Open Source. As a result, I do not support the Open Source concept as Free Software but rather as software that can grow, change and not become orphaned by the author. As has been noted in this and other lists, the cost of support is the real expense in software. The least expensive way is the method that results in the longest lifespan of the system and the lowest cost of support over the life of the product. Please do not confuse this with the cost of purchase or implementation. Lastly, what is meant by the statement: The freer things are, in most things, the more money you make and the less you pay for a unit. Who is referenced by the you. Mike --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Lastly, what is meant by the statement: The freer things are, in most things, the more money you make and the less you pay for a unit. Who is referenced by the you. That would be you in the perfect sense. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
There are a number of reasons I'm not especially interested in getting involved with GT.M development right now. I have (seriously) considered it, but I have seen little interest expressed on this list in addressing the types of issues I've attempted to call attention to, and supporting the existing code base seems to be the biggest (even only) concern of most people on this list. Greg, you have my sympathy and if i knew enough, I'd help you. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Fascinating. A language with tensed personal pronouns. --- Ruben Safir [EMAIL PROTECTED] wrote: That would be you in the perfect sense. Ruben === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
From: Ruben Safir Monday, August 22, 2005 6:50 PM I believe your confusing Free Software with the marketing speech created by Bruce Perens. It's Free Software, like in Free Market, and it has a specific definition which is detailed here http://www.gnu.org/philosophy/free-sw.html Yes, of course that is what you meant. In a Free Market, you could even sell the free software and prevent it from being distributed as Free Software by adding restrictive licensing. Mike --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
From: Ruben Safir Monday, August 22, 2005 6:52 PM Lastly, what is meant by the statement: The freer things are, in most things, the more money you make and the less you pay for a unit. Who is referenced by the you. That would be you in the perfect sense. Ruben This is perfectly ridiculous. One is either buying or selling, not both. Mike --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 18:54, Michael D. Weisner wrote: From: Ruben Safir Monday, August 22, 2005 6:50 PM I believe your confusing Free Software with the marketing speech created by Bruce Perens. It's Free Software, like in Free Market, and it has a specific definition which is detailed here http://www.gnu.org/philosophy/free-sw.html Yes, of course that is what you meant. In a Free Market, you could even sell the free software and prevent it from being distributed as Free Software by adding restrictive licensing. Well no. Strictly speaking that is called a monopoly. In fact, that is exactly the classical term for copyright and patent. Such as we have a socialist safety net in this society, we also have legal monopolies, all which diverge from a Free Market. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
This is perfectly ridiculous. One is either buying or selling, not both. It's both and everyone. Buyers, sellers, and even people who live on small islands in the pacific. Its You in the perfect sense. If you want to discuss this more, email me off list. Thanks Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 18:48, Greg Woodhouse wrote: Fascinating. A language with tensed personal pronouns. I think it is called perfect because it refers to group of people, each one individually, and at any time. It's more common in Hebrew. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
No, the issue is that it's necessary to compile MUMPS (not C). In principle, there's no reason why it can't be done. --- Ruben Safir [EMAIL PROTECTED] wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISK ? Is binary outputs embedded into the source code of GM.T? Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
So, you're serious? That really is interesting. I have read that Japanese has a group of adjectives that can exhibit tense, but I did not know that pronouns in Hebrew could exhibit aspect like this. --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 18:48, Greg Woodhouse wrote: Fascinating. A language with tensed personal pronouns. I think it is called perfect because it refers to group of people, each one individually, and at any time. It's more common in Hebrew. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
When God talks, everyone listens ;) Ruben On Mon, 2005-08-22 at 19:20, Greg Woodhouse wrote: So, you're serious? That really is interesting. I have read that Japanese has a group of adjectives that can exhibit tense, but I did not know that pronouns in Hebrew could exhibit aspect like this. --- Ruben Safir [EMAIL PROTECTED] wrote: On Mon, 2005-08-22 at 18:48, Greg Woodhouse wrote: Fascinating. A language with tensed personal pronouns. I think it is called perfect because it refers to group of people, each one individually, and at any time. It's more common in Hebrew. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 19:14, Greg Woodhouse wrote: No, the issue is that it's necessary to compile MUMPS (not C). In principle, there's no reason why it can't be done. I understand that. (or maybe I don't) But why can't GT.M compile to create binary RISK instructions for mumps with gcc? Ruben --- Ruben Safir [EMAIL PROTECTED] wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISK ? Is binary outputs embedded into the source code of GM.T? Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
Ruben wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISC ? Is binary outputs embedded into the source code of GM.T? Yes (essentially), GT.M includes a MUMPS compiler that generates machine language for the target processor that it runs on. GT.M for Linux/x86 has a virtual machine model with specifics for x86. Other GT.M implementations (not Free yet) target other processors and it might be that a model for PPC could be easily added by someone familiar with that processor. I have looked at the GT.M sources and I admire the clean modular breakdown of its design for supporting multiple types of CPU and multiple operating systems, but that type of programming is outside of my expertise and interests at the moment. As a matter of principle, I would like to help out with such a project so if someone with the expertise and time to make it happen starts to work on such a project, please let me know if you think I can help. --- Jim Self Systems Architect, Lead Developer VMTH Computer Services, UC Davis (http://www.vmth.ucdavis.edu/us/jaself) --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: GTM on OSX WAS: [Hardhats-members] more M read questions
I believe the issue was related to compiler specific optimizations in the C implementation of the M compiler. Bhaskar's been quiet lately but we've discussed this on the list before. I was interested because I simply love my MAC. Even though I'm about to paste in part of the discussion to port, I think we should just wait. Apple will switch to x86 next year and they will more than likely follow in the footsteps of Sun - by providing a run-time library that can execute linux compiled code natively. On top of that, FreeBSD so closely matches that Apple documents the differences easily enough to scope the port: http://developer.apple.com/documentation/Darwin/Conceptual/KernelProgram ming/BSD/chapter_11_section_3.html Here's a piece of our previous discussion... -- On Wed, 2005-04-13 at 01:46 -0500, chuck5566 wrote: Agree wholeheartedly, Chris. I would suggest: 1st - Determining that level of interest, and where it's at. Are people really interested in a GT.M for OS X, or would clients on OS X that could converse with GT.M and the RPC broker (on a Linux box elsewhere) be enough? Or both? Might be time for a Hardhats-OSX list. [KSB] Since there is a GT.M (non open source; non free) for IBM eServer pSeries (nee RS/6000) AIX, a port to Mac OS X from this would be straightforward, but would need to be performed by Fidelity. A port to Mac OS X from GT.M on x86 GNU/Linux (open source free) would require retargeting the M compiler (the database would just go over, since it vanilla UNIX for the most part). So, creating a client would be almost as much work as porting GT.M. 2nd - If the interest for GT.M on OS X is sufficient, I'd first straighten out the legalities before starting any work or even looking for funding. Chuck GT.M on x86 GNU/Linux is released under the GNU General Public License (GPL). If it is used to port GT.M to Mac OS X by anyone other than Fidelity, then the resulting work would be covered by the GPL, and is best released under the GPL. -- Bhaskar -- David Sommers, Architect | Dialog Medical -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Greg Woodhouse Sent: Monday, August 22, 2005 7:15 PM To: hardhats-members@lists.sourceforge.net Subject: RE: [Hardhats-members] more M read questions No, the issue is that it's necessary to compile MUMPS (not C). In principle, there's no reason why it can't be done. --- Ruben Safir [EMAIL PROTECTED] wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISK ? Is binary outputs embedded into the source code of GM.T? Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Greg said: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISK ? Is binary outputs embedded into the source code of GM.T? Ruben The GT.M compiler for the RISC architecture does generate correct machine code. As I understand it, the GT.M compiler generates an intermediate code which is stored in the .o files, and then has a Just-In-Time Native code generator to produce the machine code that is actually run. By the way, it is this Native code that bloats each user process's memory profile. If the Native Code were stored in the .o files, it would lower the memory footprint at the cost of requiring the GT.M compiler to be able to handle every operating system's native object format. We (collectively) have the code for GT.M on x86 Linux, if we don't want to just find enough economic support for Macs to convince Fidelity that it is a good investment, we could just do the port to Macs ourselves. As busy as I personally stay, I'd probably be in favour of the economic support theory, which I will support as I can, even if the community chooses the best way would be to just to have a bake sale. David Whitten (713) 870-3834 --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
Is MUMPS + VistA used on any other big endian machine? Just curious. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
Is MUMPS + VistA used on any other big endian machine? Just curious.=20 J. I don't recall if the Data General or IBM 360/370 machines were big endian. MUMPS has run on so many machines that I feel confident that it has run on as many variations of computers as you wish. No one to my knowledge has tried VistA on all of the varieties of machines supported by Cache and GT.M, however I expect it can run on all of them. MUMPS code very rarely deals with anything at the level of big-endian and little-endian. The ANSI/ISO standard pretty much is written to make it very difficult for a Standard MUMPS routine to be able to tell. In fact, a MUMPS program isn't even supposed to be able to tell if the characters used are stored internally in ASCII or EBCDIC. Handling numbers for character values greater than 127 requires a different characterset profile, so Unicode might be able to be detected. David --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 16:58 -0700, Jim Self wrote: Ruben wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISC ? Is binary outputs embedded into the source code of GM.T? Yes (essentially), GT.M includes a MUMPS compiler that generates machine language for the target processor that it runs on. GT.M for Linux/x86 has a virtual machine model with specifics for x86. Other GT.M implementations (not Free yet) target other processors and it might be that a model for PPC could be easily added by someone familiar with that processor. Does this mean its going to have trouble to run on 64 bit arch? Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
I'm fairly certain the Alpha is big endian. In fact, I think pretty much everything except Intel is big endian. === Gregory Woodhouse [EMAIL PROTECTED] It is foolish to answer a question that you do not understand. --G. Polya (How to Solve It) On Aug 22, 2005, at 7:15 PM, jae kim wrote: Is MUMPS + VistA used on any other big endian machine? Just curious. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Actually, the problems are a lot more severe than the issue of 32 vs. 64 bit architectures. MUMPS is very free in allowing execution of strings built at runtime (much like LISP). Think about how you might go about compiling LISP to native code without relying on any sort of abstract (virtual) machine. === Gregory Woodhouse [EMAIL PROTECTED] Nothing is as powerful than an idea whose time has come. -- Victor Hugo On Aug 22, 2005, at 8:30 PM, Ruben Safir wrote: On Mon, 2005-08-22 at 16:58 -0700, Jim Self wrote: Ruben wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISC ? Is binary outputs embedded into the source code of GM.T? Yes (essentially), GT.M includes a MUMPS compiler that generates machine language for the target processor that it runs on. GT.M for Linux/x86 has a virtual machine model with specifics for x86. Other GT.M implementations (not Free yet) target other processors and it might be that a model for PPC could be easily added by someone familiar with that processor. Does this mean its going to have trouble to run on 64 bit arch? Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
yes, i think so. I liked big-endian better because my Fortran code never had to see Intel chip. Later I had to write some apps to convert big-endian data (Irix) to be read in little-endian supercomputer. Just googled this: http://www.intersystems.com/cache/downloads/documentation/cache5docs/PDFS/ACVE_UsingCVEndian.pdf so, it (Cache) can be run on both big- and little-endian machines but the database file itself has to be converted first, if someone wants to read patient A's file from hospital Y's Sparc by hospital Z's Windows XP server. Somebody should write the code so converting to the different endianism (reading files from different endian machine) can be set up as a parameter at the execution time, negating the above tool. Sorry for the nagging... intersystem... Jae. On 8/22/05, Gregory Woodhouse [EMAIL PROTECTED] wrote: I'm fairly certain the Alpha is big endian. In fact, I think pretty much everything except Intel is big endian. === Gregory Woodhouse [EMAIL PROTECTED] It is foolish to answer a question that you do not understand. --G. Polya (How to Solve It) On Aug 22, 2005, at 7:15 PM, jae kim wrote: Is MUMPS + VistA used on any other big endian machine? Just curious. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
ugghh, I didn't know the other thread was talking about reading files by M. It must be the longest thread, 69 emails so far. anyway, I stopped reading it after about 20. i'll wait until all the smart people figure out how to handle millions of patients worth of data file to convert from 32 bit big-endian to 64 bit little-endian machine. j. On 8/22/05, jae kim [EMAIL PROTECTED] wrote: yes, i think so. I liked big-endian better because my Fortran code never had to see Intel chip. Later I had to write some apps to convert big-endian data (Irix) to be read in little-endian supercomputer. Just googled this: http://www.intersystems.com/cache/downloads/documentation/cache5docs/PDFS/ACVE_UsingCVEndian.pdf so, it (Cache) can be run on both big- and little-endian machines but the database file itself has to be converted first, if someone wants to read patient A's file from hospital Y's Sparc by hospital Z's Windows XP server. Somebody should write the code so converting to the different endianism (reading files from different endian machine) can be set up as a parameter at the execution time, negating the above tool. Sorry for the nagging... intersystem... Jae. On 8/22/05, Gregory Woodhouse [EMAIL PROTECTED] wrote: I'm fairly certain the Alpha is big endian. In fact, I think pretty much everything except Intel is big endian. === Gregory Woodhouse [EMAIL PROTECTED] It is foolish to answer a question that you do not understand. --G. Polya (How to Solve It) On Aug 22, 2005, at 7:15 PM, jae kim wrote: Is MUMPS + VistA used on any other big endian machine? Just curious. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
Actually, I think you may be confusing a couple of issues here (or maybe not). A 64-bit platform operates on data in 64-bit chunks at the instruction set level. When you load a value from main memory into a register, it is a 64-bit value that you load. When you add two integers, it is 64-bit integers that you add. How data is stored on disk is a separate issue. === Gregory Woodhouse [EMAIL PROTECTED] Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away. -- Antoine de Saint-Exupery On Aug 22, 2005, at 9:27 PM, jae kim wrote: ugghh, I didn't know the other thread was talking about reading files by M. It must be the longest thread, 69 emails so far. anyway, I stopped reading it after about 20. i'll wait until all the smart people figure out how to handle millions of patients worth of data file to convert from 32 bit big-endian to 64 bit little-endian machine. j. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
On Tue, 2005-08-23 at 00:44, Gregory Woodhouse wrote: Actually, I think you may be confusing a couple of issues here (or maybe not). A 64-bit platform operates on data in 64-bit chunks at the instruction set level. When you load a value from main memory into a register, it is a 64-bit value that you load. When you add two integers, it is 64-bit integers that you add. How data is stored on disk is a separate issue. Gotchya, but don't they also have a different instruction set when the code is moved and read from ram? Or is the CPU reading 8 bit instructions (or do I need to crack that assembeler book again) Ruben === Gregory Woodhouse [EMAIL PROTECTED] Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away. -- Antoine de Saint-Exupery On Aug 22, 2005, at 9:27 PM, jae kim wrote: ugghh, I didn't know the other thread was talking about reading files by M. It must be the longest thread, 69 emails so far. anyway, I stopped reading it after about 20. i'll wait until all the smart people figure out how to handle millions of patients worth of data file to convert from 32 bit big-endian to 64 bit little-endian machine. j. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
On Mon, 2005-08-22 at 23:49, Gregory Woodhouse wrote: Actually, the problems are a lot more severe than the issue of 32 vs. 64 bit architectures. MUMPS is very free in allowing execution of strings built at runtime (much like LISP). Think about how you might go about compiling LISP to native code without relying on any sort of abstract (virtual) machine. Ugh I hate the security implications of this as well. Random user code being interpreted and then executed.not cool without a ton of restraints. Ruben === Gregory Woodhouse [EMAIL PROTECTED] Nothing is as powerful than an idea whose time has come. -- Victor Hugo On Aug 22, 2005, at 8:30 PM, Ruben Safir wrote: On Mon, 2005-08-22 at 16:58 -0700, Jim Self wrote: Ruben wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISC ? Is binary outputs embedded into the source code of GM.T? Yes (essentially), GT.M includes a MUMPS compiler that generates machine language for the target processor that it runs on. GT.M for Linux/x86 has a virtual machine model with specifics for x86. Other GT.M implementations (not Free yet) target other processors and it might be that a model for PPC could be easily added by someone familiar with that processor. Does this mean its going to have trouble to run on 64 bit arch? Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
I certainly intend to stay with OS X as long as it's practical to do so, and I know that there is a contingent of clinicians out there that would very much prefer to not give up their Macs. I'd like to request a separately mailing list for us Mac fanatics so that we don't bore the snot out of the rest of you. I had started to looking at porting the RPC client code to OS X, but now I've been enchanted by the Siren Song of the Widget. :-) On Aug 22, 2005, at 6:58 PM, David Sommers wrote: I believe the issue was related to compiler specific optimizations in the C implementation of the M compiler. Bhaskar's been quiet lately but we've discussed this on the list before. I was interested because I simply love my MAC. Even though I'm about to paste in part of the discussion to port, I think we should just wait. Apple will switch to x86 next year and they will more than likely follow in the footsteps of Sun - by providing a run-time library that can execute linux compiled code natively. On top of that, FreeBSD so closely matches that Apple documents the differences easily enough to scope the port: http://developer.apple.com/documentation/Darwin/Conceptual/ KernelProgram ming/BSD/chapter_11_section_3.html Here's a piece of our previous discussion... -- On Wed, 2005-04-13 at 01:46 -0500, chuck5566 wrote: Agree wholeheartedly, Chris. I would suggest: 1st - Determining that level of interest, and where it's at. Are people really interested in a GT.M for OS X, or would clients on OS X that could converse with GT.M and the RPC broker (on a Linux box elsewhere) be enough? Or both? Might be time for a Hardhats-OSX list. [KSB] Since there is a GT.M (non open source; non free) for IBM eServer pSeries (nee RS/6000) AIX, a port to Mac OS X from this would be straightforward, but would need to be performed by Fidelity. A port to Mac OS X from GT.M on x86 GNU/Linux (open source free) would require retargeting the M compiler (the database would just go over, since it vanilla UNIX for the most part). So, creating a client would be almost as much work as porting GT.M. 2nd - If the interest for GT.M on OS X is sufficient, I'd first straighten out the legalities before starting any work or even looking for funding. Chuck GT.M on x86 GNU/Linux is released under the GNU General Public License (GPL). If it is used to port GT.M to Mac OS X by anyone other than Fidelity, then the resulting work would be covered by the GPL, and is best released under the GPL. -- Bhaskar -- David Sommers, Architect | Dialog Medical -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Greg Woodhouse Sent: Monday, August 22, 2005 7:15 PM To: hardhats-members@lists.sourceforge.net Subject: RE: [Hardhats-members] more M read questions No, the issue is that it's necessary to compile MUMPS (not C). In principle, there's no reason why it can't be done. --- Ruben Safir [EMAIL PROTECTED] wrote: Unlike GT.M, it does not generate machine language in compiling MUMPS source routines so I would expect no special surprizes due to the shift from X86 on FreeBSD to PPC on OS/X. Why can't the complier generate the correct machine code for the RISK ? Is binary outputs embedded into the source code of GM.T? Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members === Gregory Woodhouse [EMAIL PROTECTED] Design quality doesn't ensure success, but design failure can ensure failure. --Kent Beck --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development
Re: GTM on OSX WAS: [Hardhats-members] more M read questions
This will be the last email, promise. I guess my terribly unnecessary concern is that when VA hospitals with current 32 bit whatever machine ( are they using Windows on Intel now? or IBM w/ PPC?) want to upgrade their hardware to new 64 bit Intel/AMD chip, (very possible scenario in the near future) the old database file should be read without manipulation by the new platform/chip/M/VistA. Good night. J. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
[Hardhats-members] more M read questions
The read command in M seems to be the most complicated function it has. I am trying to perform a binary read. I do it this way: read blockIn#255 The problem is that as I debug the code, $length(blockIn) does not always=255. I think this is because sometimes the stream contains a terminator, such as a #13 etc. How do do a read that ignores the usual terminators? Thanks Kevin --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Kevin; There is only a single data-type in MUMPS, strings. What you are doing is a fixed length buffer read of characters (real characters or binary data). You are opening up a big bag of issues which the MDC argued over for decades. If you are talking about binary, are you talking about big-endian or little-endian representation (what do the bits mean?). By dealing in characters, we don't have to worry about byte order per word. Now some implementations did provide tools for doing these operations (most notable was Micronetics (now InterSystems). I believe that GTM has some of these same tools. They also have the thinnest binding with the underlying operating system, so poking out to do this type of operation is pretty simple in GT.M. - Original Message - From: Kevin Toppenberg [EMAIL PROTECTED] To: Hardhats Sourceforge [EMAIL PROTECTED] Sent: Sunday, August 21, 2005 6:46 AM Subject: [Hardhats-members] more M read questions The read command in M seems to be the most complicated function it has. I am trying to perform a binary read. I do it this way: read blockIn#255 The problem is that as I debug the code, $length(blockIn) does not always=255. I think this is because sometimes the stream contains a terminator, such as a #13 etc. How do do a read that ignores the usual terminators? Thanks Kevin --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Kevin, You are on the right track. Increasing the number of characters per READ is by far the most significant thing you can do to speed up your routine. Reading one character at a time using a star-Read is very slow. Each M implementation has a way to do binary reads -- ie, a read which does not look for a terminator and does not translate any characters (like HT into spaces), but the M Standard does not specify this level of detail -- it's left to the implementer. I don't know whether VistA provides a way to call a file Open that provides the necessary parameters for this. Others on this list will. Most M's do not have a problem storing binary data strings in globals. (I know of only one that uses null-terminated strings, and to my knowledge, it has never been used for VistA.) WHY do this at all? It seems like the long-way around. Normally, when a file is the object of interest, one just points to it by name and lets the underlaying OS and utilities handle it. - Original Message - From: Kevin Toppenberg [EMAIL PROTECTED] To: Hardhats Sourceforge [EMAIL PROTECTED] Sent: Sunday, August 21, 2005 8:46 AM Subject: [Hardhats-members] more M read questions The read command in M seems to be the most complicated function it has. I am trying to perform a binary read. I do it this way: read blockIn#255 The problem is that as I debug the code, $length(blockIn) does not always=255. I think this is because sometimes the stream contains a terminator, such as a #13 etc. How do do a read that ignores the usual terminators? Thanks Kevin --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. If you are going to stay strictly within ANSI standard M, then binary data is best handled outside of the M environment. This is not really a VA Kernel issue since the Kernel is adhering to ANSI M for its globals. - Original Message - From: Maury Pepper [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Sunday, August 21, 2005 12:52 PM Subject: Re: [Hardhats-members] more M read questions Kevin, You are on the right track. Increasing the number of characters per READ is by far the most significant thing you can do to speed up your routine. Reading one character at a time using a star-Read is very slow. Each M implementation has a way to do binary reads -- ie, a read which does not look for a terminator and does not translate any characters (like HT into spaces), but the M Standard does not specify this level of detail -- it's left to the implementer. I don't know whether VistA provides a way to call a file Open that provides the necessary parameters for this. Others on this list will. Most M's do not have a problem storing binary data strings in globals. (I know of only one that uses null-terminated strings, and to my knowledge, it has never been used for VistA.) WHY do this at all? It seems like the long-way around. Normally, when a file is the object of interest, one just points to it by name and lets the underlaying OS and utilities handle it. --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
On Sun, 2005-08-21 at 15:13 -0400, smcphelan wrote: Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. This is all beyond me because all data is just data. ASCI, Binary, all the same thing. No matter how you look at it, a byte can only have one of 256 representations. The rest is all interpretation. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character Then there were 36 bit words (6 (6-bit) characters per word (Univac FIELDDATA), or 10 6-bit characters/word, but each of these are mapping systems for characters. - Original Message - From: Ruben Safir [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Sunday, August 21, 2005 12:46 PM Subject: Re: [Hardhats-members] more M read questions On Sun, 2005-08-21 at 15:13 -0400, smcphelan wrote: Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. This is all beyond me because all data is just data. ASCI, Binary, all the same thing. No matter how you look at it, a byte can only have one of 256 representations. The rest is all interpretation. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
On Sun, 2005-08-21 at 12:57 -0700, Chris Richardson wrote: Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character In terms of storing and retrieving data, it shouldn't matter. We've been dealing with 7 bit encoded attachments for a decade in email. Ruben Then there were 36 bit words (6 (6-bit) characters per word (Univac FIELDDATA), or 10 6-bit characters/word, but each of these are mapping systems for characters. - Original Message - From: Ruben Safir [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Sunday, August 21, 2005 12:46 PM Subject: Re: [Hardhats-members] more M read questions On Sun, 2005-08-21 at 15:13 -0400, smcphelan wrote: Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. This is all beyond me because all data is just data. ASCI, Binary, all the same thing. No matter how you look at it, a byte can only have one of 256 representations. The rest is all interpretation. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
RE: [Hardhats-members] more M read questions
Kevin, it may be easier to encode the binary into a subset of ASCII/ANSI that is supported by M string. There are many definitions on what a string *is* depending on the language and system - but I'm sure you can find a codeset that fits. Base64 in the worse case. /David. David Sommers, Architect | Dialog Medical -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ruben Safir Sent: Sunday, August 21, 2005 3:47 PM To: [EMAIL PROTECTED] Subject: Re: [Hardhats-members] more M read questions On Sun, 2005-08-21 at 15:13 -0400, smcphelan wrote: Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. This is all beyond me because all data is just data. ASCI, Binary, all the same thing. No matter how you look at it, a byte can only have one of 256 representations. The rest is all interpretation. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
Let's be practical. There seem to be only a few M environments. Are any of them using 6 bit bytes etc? Do any of the underlying file systems server other than an 8 bit byte when asked to read one unit (byte) from a file? Yes, there are widecharacter strings, but the underlying filesystem still deals with them as 8-bit bytes, doesn't it? Kevin On 8/21/05, Chris Richardson [EMAIL PROTECTED] wrote: Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character Then there were 36 bit words (6 (6-bit) characters per word (Univac FIELDDATA), or 10 6-bit characters/word, but each of these are mapping systems for characters. - Original Message - From: Ruben Safir [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Sunday, August 21, 2005 12:46 PM Subject: Re: [Hardhats-members] more M read questions On Sun, 2005-08-21 at 15:13 -0400, smcphelan wrote: Here, here. Chris also stated this. ANSI standard M is not really designed to handle binary data. This is one reason Intersystems added extensions (if you wish to call it that). But then you are bound to a specific M vendor's implementation, in this case, Cache. This is all beyond me because all data is just data. ASCI, Binary, all the same thing. No matter how you look at it, a byte can only have one of 256 representations. The rest is all interpretation. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members
Re: [Hardhats-members] more M read questions
The key word (not the keyword) here is encoded. Fileman actually provides utilities for hexadecimal encoding that I've found useful on more than one occasion. There are really two issues here: whether the M implementation can handle binary data (not necessarily), and whether applications can work with binary data. In the latter case, there is the problem that M programmers have a propensity for using sentinel values like ^ to delimit data items, but the first problem is much more serious. === Gregory Woodhouse [EMAIL PROTECTED] The most incomprehensible thing about the world is that it is at all comprehensible. --Albert Einstein (1879-1955) On Aug 21, 2005, at 1:26 PM, Ruben Safir wrote: On Sun, 2005-08-21 at 12:57 -0700, Chris Richardson wrote: Ah, but how big is a character? MUMPS deals in characters reguardless of the number of octets required to represent it. 1Octet = 8 bits Ascii - 1 octet/character Unicode, Kanji,Katakana,etc - 2 octets/character ISO-10646 - 4octets/character In terms of storing and retrieving data, it shouldn't matter. We've been dealing with 7 bit encoded attachments for a decade in email. Ruben --- SF.Net email is Sponsored by the Better Software Conference EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile Plan-Driven Development * Managing Projects Teams * Testing QA Security * Process Improvement Measurement * http://www.sqe.com/bsce5sf ___ Hardhats-members mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/hardhats-members