
                  OS/2 Programming                 (Fidonet)

                  Sunday, 26-Sep-1999 to Friday, 01-Oct-1999

+----------------------------------------------------------------------------+

From: Mario Semo                                        10-Sep-99 08:17:27
  To: Ian Moote                                         28-Sep-99 18:28:06
Subj: Maximum threads.

Hallo Ian!

Antwort auf eine Message von Ian Moote an MIKE RUSKAI:

 IM> Most of the threads will be identical and will just be running
 IM> different  instance data, so process space shouldn't be a problem.
 IM> And with what  i've got running that gives me probably just under
 IM> 4000 threads to play  with. _Should_ be more than enough.

if you run another app which eats 3000 threads, you will just be able to run
1000. oh, and some threads are in the OS itself :-). 

Servus, Mario!

--- FleetStreet 1.21 PR#2
 * Origin: LC/32 Development Team, http://www.kirchnersoft.com (2:310/14.11)

+----------------------------------------------------------------------------+

From: Mario Semo                                        10-Sep-99 08:19:24
  To: Ian Moote                                         28-Sep-99 18:28:06
Subj: Maximum threads.

Hallo Ian!

Antwort auf eine Message von Ian Moote an HERBERT ROSENAU:

 HR>> So you can set THREADS=4095. But not 4096 or higher.

just as a note.

after installing VACPP 3.0 (as i remember, or maybe it was 2.1, but i think
3.0) you have a 
SET THREADS=1024
in your config.sys file. but this statement is just for fun. and eat some
environment space.
the correct statement should be
THREADS=1024
os2 installation correctly sets this value (but to a lower value, depends on
OS2s version you have)


Servus, Mario!

--- FleetStreet 1.21 PR#2
 * Origin: LC/32 Development Team, http://www.kirchnersoft.com (2:310/14.11)

+----------------------------------------------------------------------------+

From: Tom Torfs                                         28-Sep-99 17:07:00
  To: Thomas Seeling                                    01-Oct-99 05:53:17
Subj: ... Perl

Thomas Seeling wrote in a message to Tom Torfs:

 TT>>          c = (((c-'a')+13)%26)+'a';
 TT> This assumes contiguous letters of course.

 TS> afair Ansi-/ISO-C forces letters and digits to be ordinally
 TS> contiguous. 

You remembered wrong. Only the digits have such a guarantee. For letters this
would not be feasible, since in the second most used family of character sets
(EBCDIC) this isn't the case.

2.2.1 Character sets (this is 5.2.1 in ISO numbering)

<snip>

the 10 decimal digits 

         0  1  2  3  4  5  6  7  8  9

<snip>

In both the source and execution basic character sets, the value of each
character after 0 in the above list of decimal digits shall be one greater
than the value of the previous.

greetings,
Tom
tomtorfs@village.uunet.be

--- timEd/2 1.10.y2k+
 * Origin: 80X86 BBS 32-15-24.62.32 V.34/V.FC (24h/24h) (2:292/516)

+----------------------------------------------------------------------------+

From: Herbert Rosenau                                   30-Sep-99 04:21:10
  To: Thomas Seeling                                    01-Oct-99 05:53:17
Subj: ... Perl

 TS> afair Ansi-/ISO-C forces letters and digits to be ordinally
 TS> contiguous.

No. The only garanteed is: digits are contigous. So '0' through '9' can be
simple in/decremented. But letters not.

But practise thows tha independant of chasets and real computers it would go.
And if you don't mix numbers and letters (like 0x41 for 'A') and uppercase and 
lowercase it would be independant of ASCII or EBCDIC.

But the standard can not and will not guaratee that for all computers and each 
environment your program my run.

--- Sqed/32 1.14/development
 * Origin: Schont die Umwelt: Vermeidet DOSen (2:2476/493)
102
2320/38

+----------------------------------------------------------------------------+

From: Tobias Ernst                                      30-Sep-99 10:49:05
  To: Thomas Seeling                                    01-Oct-99 05:53:17
Subj: Letters are not assumed to be contiguous!

Hallo Thomas!

 TT>>>          c = (((c-'a')+13)%26)+'a';
 TT>> This assumes contiguous letters of course.

 TS> afair Ansi-/ISO-C forces letters and digits to be ordinally 
 TS> contiguous.

No! Digits are forced to be contiguous, but letters are not. That is why
ctype.h provides macros like toupper and tolower, and isalpha. Compilers for
ASCII platforms may well implement those macros in the way above, but the
programmer should never do it.

The reason is obvious: All existing charsets have contiguous digits, but not
all of them have contiguous digits. For example EBCDIC (used in S/390 and
AS/400) does not have contiguous letters. EBCDIC has 0..F (hex digits 0..9,
a..f) contiguos, but the other letters were added at a later time and
therefore follow at another position.

And: OS/390 uses EBCDIC, and does not only have an ISO-certified C compiler,
but, when using the Open Edition, it even has all POSIX certificates to allow
itself to be called "Unix". 

Thinking about it, implementing toupper like upper = (lower -'a') + 'A' might
even work in EBCDIC, because the displacemenet between f and g is the same as
that between F and G, but implementing isalpha like  

  (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')

certainly would not work on EBCDIC. This is a major issue in porting spaghetti 
code applications like R/3 <grin> to OS/390.

Viele Gre,
Tobias

--- Msged/BSD TE 06 (pre)
 * Origin: Running FreeBSD 3.2 (2:2476/418)
102
2320/38

+----------------------------------------------------------------------------+

From: mark lewis                                        29-Sep-99 10:18:00
  To: David Noon                                        01-Oct-99 05:53:17
Subj: Did you get a Netmail?

 dn> I sent you a netmail a couple of months ago regarding this
 dn> echo and its
 dn> e-list entry. I have not received a response so I suspect
 dn> netmail is broken.
 dn> [What's new there?]

i did get a netmail and did reply... am working from "extreme" remote ATM...

 dn> Can you please reply to dwnoon@ibm.net?

will do so shortly... have to get back to home base later today or tomorrow... 
will make all attempts to handle this then...

)\/(ark

 * Origin:  (1:3634/12)

+----------------------------------------------------------------------------+

+============================================================================+
