ann: project (small-c) pcdoslib and CCNA.EXE
s_dub...@yahoo.com
small-c in a pcdos enviornment. The pcdoslib consists of the minimal
list of interface functions to effect a small-c self-compile, using
the pcdos int 21h file handle functions.
found here: http://www.project-fbin.hostoi.com/index.htm
Steve
Rod Pemberton
news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
> I've done a pcdoslib for small-c for those who would be interested in
> small-c in a pcdos enviornment. The pcdoslib consists of the minimal
> list of interface functions to effect a small-c self-compile, using
> the pcdos int 21h file handle functions.
>
>
FYI, I cannot post correct links to your site through AIOE.org, even in
reply. AIOE.org uses URIBL.com's banlists to prevent spam. URIBL.com bans
the domain your site is on because someone is sending email spam from that
domain in violation of 000webhost's policies. I posted a complaint with
000webhost a while back, and they got URIBL fixed, for about a month ...
So, if you have a problem posting a link to your site somewhere or with
email from your host rejected as spam, that's what's going on. If it
happens to you, like when posting to a forum, you could use tinyurl to post
a different link. I'm not sure what you'd do if you free webhost email is
being blacklisted, maybe use another email service ...
> ... pcdos int 21h file handle functions ...
My initial response to reading that is a happy one!
My experiments:
1) (yesterday)
in Win98SE DOS console window
converted PCDOSLIB.C and ITOA.C to ANSI parameters
compiled with the version of SmallC I've been working on
attempted to see if it was a drop in replacement for C5LIB.C
compiler hangs when compiling
2)
in Win98SE DOS console window
built CCNA.COM using CCNA.MAK
compiler hangs when compiling
Sigh... So, I decided to reboot into RM MS-DOS to see if it worked:
3)
reboot into MS-DOS v7.10 in RM
didn't rebuild
SHOCK: compiler doesn't hang when compiling!
Unfortunately, it is exceptionally slow compared to earlier versions. On my
machine, "my" SmallC takes 2 to 5 seconds to compile. I thought your early
versions were that quick too. This one took 10 minutes or more to compile
itself. It was taking from 2 to 30 seconds *per* procedure ... Also, the
emitted assembly and compiled .BIN for a 1st gen. recompile of CCNA is
_different_ from the .NSM and .BIN for CCNA respectively. The first 4K or
so seems to be the same, from there ... It was taking too long to compile.
So, I haven't yet done another recompile stage to see if it was identical.
Since it compiled, I decided to see if it'd work as a replacement for C5LIB,
again ...
4)
a)Win98SE DOS console window
compile "my" SmallC with PCDOSLIB.C and ITOA.C from C5LIB version
b)reboot in MS-DOS v7.10 in RM
compiler hangs when compiling
So, PCDOSLIB.C does work for me, but only in RM MS-DOS and inconjuction with
CCNA. CCNA or PCDOSLIB or something else was extremely slow, maybe some of
those other new C routines ... PCDOSLIB, AFAICT, cannot directly replace
C5LIB.C yet. I.e., I think there are some dependencies on CCNA or related
code, e.g., prolog/epilog, that may be causing the hangs in RM DOS. Since
PCDOSLIB works with CCNA, it's definately close. I found it somewhat
strange that CCNA compiled with PCDOSLIB works in RM but hangs in a Win98SE
DOS console. If it's a normal DOS program, I'd think it'd work for both.
I haven't looked at the code yet, but if I see or find something, I'll keep
you posted.
Rod Pemberton
s_dub...@yahoo.com
wrote:
> <s_dubrov...@yahoo.com> wrote in message
>
> news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
>
> > I've done a pcdoslib for small-c for those who would be interested in
> > small-c in a pcdos enviornment. The pcdoslib consists of the minimal
> > list of interface functions to effect a small-c self-compile, using
> > the pcdos int 21h file handle functions.
>
> > found here: [snip]
>
[partial quick response follows..]
> FYI, I cannot post correct links to your site through AIOE.org, even in
> reply. AIOE.org uses URIBL.com's banlists to prevent spam. URIBL.com bans
> the domain your site is on because someone is sending email spam from that
> domain in violation of 000webhost's policies. I posted a complaint with
> 000webhost a while back, and they got URIBL fixed, for about a month ...
> So, if you have a problem posting a link to your site somewhere or with
> email from your host rejected as spam, that's what's going on. If it
> happens to you, like when posting to a forum, you could use tinyurl to post
> a different link. I'm not sure what you'd do if you free webhost email is
> being blacklisted, maybe use another email service ...
>
Thanks, I didn't know that is occurring..
> > ... pcdos int 21h file handle functions ...
>
> My initial response to reading that is a happy one!
>
> My experiments:
>
> 1) (yesterday)
> in Win98SE DOS console window
> converted PCDOSLIB.C and ITOA.C to ANSI parameters
> compiled with the version of SmallC I've been working on
> attempted to see if it was a drop in replacement for C5LIB.C
> compiler hangs when compiling
>
Hmm, I'll see if I can check win98se, I don't currently have the
win98se hard drive installed on a system.
> 2)
> in Win98SE DOS console window
> built CCNA.COM using CCNA.MAK
> compiler hangs when compiling
>
> Sigh... So, I decided to reboot into RM MS-DOS to see if it worked:
>
> 3)
> reboot into MS-DOS v7.10 in RM
> didn't rebuild
> SHOCK: compiler doesn't hang when compiling!
>
I wonder what version of dos that Win98SE DOS console window reports,
to a dos program?
> Unfortunately, it is exceptionally slow compared to earlier versions. On my
> machine, "my" SmallC takes 2 to 5 seconds to compile. I thought your early
> versions were that quick too. This one took 10 minutes or more to compile
> itself. It was taking from 2 to 30 seconds *per* procedure ... Also, the
> emitted assembly and compiled .BIN for a 1st gen. recompile of CCNA is
> _different_ from the .NSM and .BIN for CCNA respectively. The first 4K or
> so seems to be the same, from there ... It was taking too long to compile.
> So, I haven't yet done another recompile stage to see if it was identical.
> Since it compiled, I decided to see if it'd work as a replacement for C5LIB,
> again ...
>
Yes, it is a couple of orders of magnatude slower on XP in cmd.exe
too. It's the ABI interface to the Int21h calls. More information
needs to be saved on return than there are registers (at least on the
initial first glance, I should rethink this now that the function set
is known) so memory variables are used to snapshot the general purpose
registers, so this is much slower, esp. for character i/o. Call5 abi,
for historical reasons, limits the abi to 2 parameters: 'function
number' and 'information address', so these are easily implemented by
passing on the stack. The handle functions, as a group, use all of
the GP regs, plus AX returned has different meaning depending on the
CF state (error or not) so the CF state also needs to be passed back
to the 'C' fn caller, such as in detecting EOF vs int chr, returned.
I guess the more use ccna receives, the more drive to optimize this.
> 4)
> a)Win98SE DOS console window
> compile "my" SmallC with PCDOSLIB.C and ITOA.C from C5LIB version
> b)reboot in MS-DOS v7.10 in RM
> compiler hangs when compiling
>
> So, PCDOSLIB.C does work for me, but only in RM MS-DOS and inconjuction with
> CCNA. CCNA or PCDOSLIB or something else was extremely slow, maybe some of
> those other new C routines ... PCDOSLIB, AFAICT, cannot directly replace
> C5LIB.C yet. I.e., I think there are some dependencies on CCNA or related
> code, e.g., prolog/epilog, that may be causing the hangs in RM DOS. Since
> PCDOSLIB works with CCNA, it's definately close. I found it somewhat
> strange that CCNA compiled with PCDOSLIB works in RM but hangs in a Win98SE
> DOS console. If it's a normal DOS program, I'd think it'd work for both.
>
Yeah, but you are at the mercy of how well the emulation of Dos
console mimics native dos. IIRC, win98 had its issues on certain
edges.
The open/create handle function [int 21h, 6Ch], according to my
footnotes, is supported in dos v.4.0+. Possibly, that is an issue
here.
> I haven't looked at the code yet, but if I see or find something, I'll keep
> you posted.
>
Thanks alot for your detail and quick attention, very helpful.
..to be continued..
Steve
> Rod Pemberton
s_dub...@yahoo.com
wrote:
I guess you meant ccna.exe? This prolog generates an exe header.
Anyway, I assembled ccna.nsm on a win95 DOS console window system to
ccna.exe and used that to compile ccna.c into ccnax.nsm, and it is
taking 3 minutes to self-compile. So now I know how to make blazing
fast processors crawl ;-)
> Sigh... So, I decided to reboot into RM MS-DOS to see if it worked:
>
> 3)
> reboot into MS-DOS v7.10 in RM
> didn't rebuild
> SHOCK: compiler doesn't hang when compiling!
>
> Unfortunately, it is exceptionally slow compared to earlier versions. On my
> machine, "my" SmallC takes 2 to 5 seconds to compile. I thought your early
> versions were that quick too. This one took 10 minutes or more to compile
> itself. It was taking from 2 to 30 seconds *per* procedure ... Also, the
> emitted assembly and compiled .BIN for a 1st gen. recompile of CCNA is
> _different_ from the .NSM and .BIN for CCNA respectively. The first 4K or
> so seems to be the same, from there ... It was taking too long to compile.
Perhaps one of the include files was missing.
> So, I haven't yet done another recompile stage to see if it was identical.
> Since it compiled, I decided to see if it'd work as a replacement for C5LIB,
> again ...
>
> 4)
> a)Win98SE DOS console window
> compile "my" SmallC with PCDOSLIB.C and ITOA.C from C5LIB version
> b)reboot in MS-DOS v7.10 in RM
> compiler hangs when compiling
>
Does 'your' small-c use BX as primary, and DX as secondary registers?
ccna does.
> So, PCDOSLIB.C does work for me, but only in RM MS-DOS and inconjuction with
> CCNA. CCNA or PCDOSLIB or something else was extremely slow, maybe some of
> those other new C routines ... PCDOSLIB, AFAICT, cannot directly replace
> C5LIB.C yet. I.e., I think there are some dependencies on CCNA or related
> code, e.g., prolog/epilog, that may be causing the hangs in RM DOS. Since
> PCDOSLIB works with CCNA, it's definately close. I found it somewhat
> strange that CCNA compiled with PCDOSLIB works in RM but hangs in a Win98SE
> DOS console. If it's a normal DOS program, I'd think it'd work for both.
>
> I haven't looked at the code yet, but if I see or find something, I'll keep
> you posted.
>
Perhaps this weekend I can try a win98 platform.
Each OS's DOS console window behaves slightly differently. On the
win95 system I had to put the nasm commandline (nasm.exe -...@ccna.mak)
into a .bat batch file to assemble it. My attempts to run nasm by way
of a 'shortcut' failed. -also the output of ccna in the ask() routine
is staggered for ea. newline (LF but no CR) whereas XP's DOS console
window will CR it.
Steve
> Rod Pemberton
Rod Pemberton
news:bef0a6b9-1402-4b7b...@m9g2000yqe.googlegroups.com...
> On Jun 22, 5:23 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> wrote:
> > <s_dubrov...@yahoo.com> wrote in message
> > news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
>
>
> I guess you meant ccna.exe?
CCNA.BIN actually ...
> This prolog generates an exe header.
Sorry, I renamed it to .COM without checking it. It is an .EXE. DOS
(MS-DOS) will execute either without regards to correct naming. I.e., a
.COM can be named a .EXE and vice-versa. If MZ is present, then .exe,
else .com ...
> [... ] is taking 3 minutes to self-compile.
> So now I know how to make blazing fast processors crawl ;-)
:-)
> > It was taking too long to compile.
>
> Perhaps one of the include files was missing.
I have these from the website (html saved via browser as text):
CCNA.C
CCNA.NSM
EPILOG.NSM
INCLUDE.H
itoa.c
PCDOSLIB.C
PROLOG.NSM
RELAT.C
strcmp.c
CCNA_st.WPD
CCNA.MAK
tstmain.c
tstmain.mak
tstmain_stat.
These I believe were created:
CCNA.WPD
CCNA.BIN
CCNA.LST
CURRFILE.MAP
It seems the LFNs are lowercased on a few. The SFNs seem fine. Maybe I'll
rename as uppercase anyway, just in case that's an issue ... Off hand, I
don't recall lowercase code in the version I was using. Case mismatch? No,
it seems itoa.c and strcmp.c are included as lowercased in CCNA.C ...
Well, I uppercased them anyway.
> Does 'your' small-c use BX as primary, and DX as secondary
> registers? ccna does.
IIRC, you asked this previously. AIR, it was in a different thread and
probably for a different reason ... (?)
So, no, it uses AX (primary) and BX. It works with your (original) C5LIB.C.
I haven't looked at it yet (PCDOSLIB). You think that's it? I can change
that up ...
> On the win95 system I had to put the nasm commandline
> (nasm.exe -...@ccna.mak) into a .bat batch file to assemble it.
Really? :-) Fortunately, Win95 was before my switch to x86 platform ...
> is staggered for ea. newline (LF but no CR) whereas XP's DOS console
> window will CR it.
Yeah, I've mentioned that ...
Does XP accept CR LF? DOS wants CR LF. *nix and Posix are LF only. IIRC,
Mac's are the only platform that uses CR only. I get the staggered text
problem with DJGPP too. It's posix, so it outputs LF only. So, I wrote a
small TSR to fix that. I'm a bit surprised XP accepts CR only. Could that
be a hint that MS programmer's are using some FOSS software internally?
Rod Pemberton
s_dub...@yahoo.com
wrote:
> <s_dubrov...@yahoo.com> wrote in message
>
> news:bef0a6b9-1402-4b7b...@m9g2000yqe.googlegroups.com...
>
> > On Jun 22, 5:23 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > wrote:
> > > <s_dubrov...@yahoo.com> wrote in message
> > >news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
>
> > > built CCNA.COM using CCNA.MAK
>
> > I guess you meant ccna.exe?
>
> CCNA.BIN actually ...
>
> > This prolog generates an exe header.
>
> Sorry, I renamed it to .COM without checking it. It is an .EXE. DOS
> (MS-DOS) will execute either without regards to correct naming. I.e., a
> .COM can be named a .EXE and vice-versa. If MZ is present, then .exe,
> else .com ...
>
dosbox)emulates DOS perfectly. dosbox of XP won't load a .com > 64k,
DOS will. dosbox of XP accepts CR,LF as expected, but also treats
singular LF as CR,LF of DOS.
For a different problem of compile problems on DOS proper.
> So, no, it uses AX (primary) and BX. It works with your (original) C5LIB.C.
Which uses AX as primary..
> I haven't looked at it yet (PCDOSLIB). You think that's it? I can change
> that up ...
>
PCDOSLIB functions 'return (value);' - value returned in BX. ..so
yes.
> > On the win95 system I had to put the nasm commandline
> > (nasm.exe -...@ccna.mak) into a .bat batch file to assemble it.
>
> Really? :-) Fortunately, Win95 was before my switch to x86 platform ...
>
Yes, fortunately for you..
> > is staggered for ea. newline (LF but no CR) whereas XP's DOS console
> > window will CR it.
>
> Yeah, I've mentioned that ...
>
> Does XP accept CR LF? DOS wants CR LF. *nix and Posix are LF only. IIRC,
> Mac's are the only platform that uses CR only. I get the staggered text
> problem with DJGPP too. It's posix, so it outputs LF only. So, I wrote a
> small TSR to fix that. I'm a bit surprised XP accepts CR only. Could that
> be a hint that MS programmer's are using some FOSS software internally?
>
XP dosbox treats both CR,LF and LF the same, (*nix friendly, unlike
earlier dosboxen).
Steve
> Rod Pemberton
Rod Pemberton
news:e074e72c-456b-423e...@a11g2000yqm.googlegroups.com...
> On Jun 24, 3:33 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> wrote:
> > <s_dubrov...@yahoo.com> wrote in message
> > news:bef0a6b9-1402-4b7b...@m9g2000yqe.googlegroups.com...
> > > On Jun 22, 5:23 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > > wrote:
> > > > <s_dubrov...@yahoo.com> wrote in message
> > >
>news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
>
> dosbox of XP won't load a .com > 64k, DOS will.
Are you sure about that?
IIRC, the limit is right around 64KB somewhere. Let me see if I can track
that down before I post this ... I think I posted it a few years ago. Ok,
I stated this some time ago:
"DOS limits a .com to 0feffh (65280) bytes. That's 0ffffh (64KB) - 100h
(256) for PSP. IIRC, MS-DOS won't load a .com even if it's size is close
to 0feffh."
Ed Beroset claims this changed around Win95 or due to it:
http://groups.google.com/group/alt.lang.asm/msg/c9b50dad3761b942
Dave Dunfield says MS-DOS 5.0 had the limit, actually 65279 or 65278, but
other DOS' of the time period didn't ...
http://groups.google.com/group/comp.os.msdos.programmer/msg/4f8ab54165a0512e
> > > > On the win95 system I had to put the nasm commandline
> > > (nasm.exe [blah]) into a .bat batch file to assemble it.
>
> > Really? :-) Fortunately, Win95 was before my switch to x86 platform ...
>
> Yes, fortunately for you..
>
Well, actually, come to think of it, I don't recall exactly when I switched
... I used DOS prior to Win98, but I seem to be having a hard time
remembering which year I got my first x86 machine. I think it was '93, but
it could've been a few after that. I guess it just wan't an important
event.
Rod Pemberton
s_dub...@yahoo.com
wrote:
> <s_dubrov...@yahoo.com> wrote in message
>
> news:e074e72c-456b-423e...@a11g2000yqm.googlegroups.com...> On Jun 24, 3:33 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > wrote:
> > > <s_dubrov...@yahoo.com> wrote in message
> > >news:bef0a6b9-1402-4b7b...@m9g2000yqe.googlegroups.com...
> > > > On Jun 22, 5:23 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > > > wrote:
> > > > > <s_dubrov...@yahoo.com> wrote in message
>
> >news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
>
> ...
>
> > dosbox of XP won't load a .com > 64k, DOS will.
>
> Are you sure about that?
>
Pretty sure, I don't trust my memory much anymore but IIRC in testing
nasm -f bin and managing segments internally, I made a >64k .com with
separate CS,DS,SS,ES whose combined sizes were > 64k. So I agree with
what Ed Beroset said in the link you posted.
> IIRC, the limit is right around 64KB somewhere. Let me see if I can track
> that down before I post this ... I think I posted it a few years ago. Ok,
> I stated this some time ago:
>
> "DOS limits a .com to 0feffh (65280) bytes. That's 0ffffh (64KB) - 100h
> (256) for PSP. IIRC, MS-DOS won't load a .com even if it's size is close
> to 0feffh."
>
Well it's acedemic, test it to see. My understanding about the loader
of .com is that it just reads a .com into memory until the physical
end of the file is reached, irrespective of size. I know that XP
dosbox has the 64k limit because I tried a >64k .com load, it gives a
memory exceeded error message. That is when I started to explore -f
bin for .exe's.
> Ed Beroset claims this changed around Win95 or due to it:http://groups.google.com/group/alt.lang.asm/msg/c9b50dad3761b942
>
> Dave Dunfield says MS-DOS 5.0 had the limit, actually 65279 or 65278, but
> other DOS' of the time period didn't ...http://groups.google.com/group/comp.os.msdos.programmer/msg/4f8ab5416...
>
> > > > > On the win95 system I had to put the nasm commandline
> > > > (nasm.exe [blah]) into a .bat batch file to assemble it.
>
> > > Really? :-) Fortunately, Win95 was before my switch to x86 platform ...
>
> > Yes, fortunately for you..
>
> Well, actually, come to think of it, I don't recall exactly when I switched
> ... I used DOS prior to Win98, but I seem to be having a hard time
> remembering which year I got my first x86 machine. I think it was '93, but
> it could've been a few after that. I guess it just wan't an important
> event.
>
IIRC the first msdos I used was version 3.1 in about 1989 on an XT
clone, 8086, sometime later I added a scsi hd and installed windows
3.0 on it, which didn't see much use frankly. I had better software
packages on the z80 based amstrad pcw8256 with its graphics printer.
Getting programing info about windows 3.0 back then was tough to
impossible.
Anyway, I tried the self compile of ccna on win98, which was
successful but very slow, 5-6 minutes to do. Why is win98 twice as
slow as win95 (3 minutes)!
I remember you had posted a link to the bcompiler, and I had put that
on my debian gnu-linux box and compiled the various incremental
bootstraps with success. I may use the int 80h routines from it to
try a linuxlib version of small-c32, well see..
Steve
> Rod Pemberton
s_dub...@yahoo.com
wrote:
> <s_dubrov...@yahoo.com> wrote in message
>
> news:e074e72c-456b-423e...@a11g2000yqm.googlegroups.com...> On Jun 24, 3:33 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > wrote:
> > > <s_dubrov...@yahoo.com> wrote in message
> > >news:bef0a6b9-1402-4b7b...@m9g2000yqe.googlegroups.com...
> > > > On Jun 22, 5:23 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > > > wrote:
> > > > > <s_dubrov...@yahoo.com> wrote in message
>
> >news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
>
I guess my post of yesterday got lost in the cloud..
> ...
>
> > dosbox of XP won't load a .com > 64k, DOS will.
>
> Are you sure about that?
Yeah, pretty sure, IIRC.
It's an academic point, the thing to do it test it for your situation.
When I was testing nasm's -f bin to build a .com > 64k, I noticed the
behavior, so I began to check out .exe generation. Launching such a
thing under pure dos worked, XP's cmd.exe gives a memory error; 'out
of memory' or 'insufficient memory' - as I recall. The same
executable built as an .exe is allowed to run.
>
> IIRC, the limit is right around 64KB somewhere. Let me see if I can track
> that down before I post this ... I think I posted it a few years ago. Ok,
> I stated this some time ago:
>
> "DOS limits a .com to 0feffh (65280) bytes. That's 0ffffh (64KB) - 100h
> (256) for PSP. IIRC, MS-DOS won't load a .com even if it's size is close
> to 0feffh."
>
> Ed Beroset claims this changed around Win95 or due to it:http://groups.google.com/group/alt.lang.asm/msg/c9b50dad3761b942
>
I'm in agreement with Ray Moon whom Ed is responding to.
> Dave Dunfield says MS-DOS 5.0 had the limit, actually 65279 or 65278, but
> other DOS' of the time period didn't ...http://groups.google.com/group/comp.os.msdos.programmer/msg/4f8ab5416...
>
> > > > > On the win95 system I had to put the nasm commandline
> > > > (nasm.exe [blah]) into a .bat batch file to assemble it.
>
> > > Really? :-) Fortunately, Win95 was before my switch to x86 platform ...
>
> > Yes, fortunately for you..
>
> Well, actually, come to think of it, I don't recall exactly when I switched
> ... I used DOS prior to Win98, but I seem to be having a hard time
> remembering which year I got my first x86 machine. I think it was '93, but
> it could've been a few after that. I guess it just wan't an important
> event.
>
The first msdos I used was version 3.1, I guess around 1989 when I got
my xt clone. I put windows 3.0 on it a year or two later. Even then
I prefered my old Amstrad PCW9256 (z80) with its graphics printer and
superior text processing program, and connectivity software to connect
to a modem and to call in to BBS's of the day.
BTW, I self-compiled ccna on win98 which took about 10-11 minutes,
quite the dog but it works.
Steve
> Rod Pemberton
Rod Pemberton
news:58dd425b-1209-4e3e...@k16g2000yqm.googlegroups.com...
>
> I guess my post of yesterday got lost in the cloud..
>
Maybe. It came across on two of the better non-commercial hosts. Except
for Google Groups, I don't have a commercial host to check. What I see are
that here are nine posts in the thread so far. Google Groups has seven
posts currently. I.e., these last two from you are missing so far...
Hopefully, mine makes it. Sometimes, it takes a day or two for GG to get
caught up. They seem to have a periodic "glitch", perhaps end-of-month
archiving...? Othertimes, the post gets lost if you don't see it there at
the same time as elsewhere.
So, I didn't reply to your last post. Also, I haven't got around to fixing
the registers yet. I will.
Rod Pemberton
s_dub...@yahoo.com
Ugh, third posting attempt..
<s_dubrov...@yahoo.com> wrote in message
news:e074e72c-456b-423e...@a11g2000yqm.googlegroups.com...
> On Jun 24, 3:33 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> wrote:
> > <s_dubrov...@yahoo.com> wrote in message
> > news:bef0a6b9-1402-4b7b...@m9g2000yqe.googlegroups.com...
> > > On Jun 22, 5:23 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > > wrote:
> > > > <s_dubrov...@yahoo.com> wrote in message
>news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
...
> dosbox of XP won't load a .com > 64k, DOS will.
Are you sure about that?
! Yeah, pretty sure..
! It academic, the best thing to do is to test it for your situation &
enviornment.
! I ran across this issue while testing nasm -f bin for .com with
internal management of sections and segment register setups.
IIRC, the limit is right around 64KB somewhere. Let me see if I can
track
that down before I post this ... I think I posted it a few years
ago. Ok,
I stated this some time ago:
"DOS limits a .com to 0feffh (65280) bytes. That's 0ffffh (64KB) -
100h
(256) for PSP. IIRC, MS-DOS won't load a .com even if it's size is
close
to 0feffh."
Ed Beroset claims this changed around Win95 or due to it:
http://groups.google.com/group/alt.lang.asm/msg/c9b50dad3761b942
! I'm in agreement with what Ray Moon said..
Dave Dunfield says MS-DOS 5.0 had the limit, actually 65279 or 65278,
but
other DOS' of the time period didn't ...
http://groups.google.com/group/comp.os.msdos.programmer/msg/4f8ab5416...
> > > > On the win95 system I had to put the nasm commandline
> > > (nasm.exe [blah]) into a .bat batch file to assemble it.
> > Really? :-) Fortunately, Win95 was before my switch to x86 platform ...
! BTW, self-compile on win98 took nearly 11 minutes.
! Steve
> Yes, fortunately for you..
Well, actually, come to think of it, I don't recall exactly when I
switched
... I used DOS prior to Win98, but I seem to be having a hard time
remembering which year I got my first x86 machine. I think it was
'93, but
it could've been a few after that. I guess it just wan't an
important
event.
Rod Pemberton
Rod Pemberton
news:iueuta$v2a$1...@speranza.aioe.org...
> <s_dub...@yahoo.com> wrote in message
> news:58dd425b-1209-4e3e...@k16g2000yqm.googlegroups.com...
> >
>
> fixing the registers yet. I will.
>
I changed the retrun registers in OSfn_() for AX and BX. So, instead of
MOV BX,AX
it's now:
MOV BX, DX
Hopefully, that's correct for AX and BX.
However, I've run into another issue. I was ignoring an error parsing a
comment, since it was "just" a comment... Apparently, that wasn't a good
idea. It stopped parsing from there. After that was fixed, it seems the
version of SmallC I've got cannot parse some of the C code in PCDOSLIB. It
treats any return() as the end-of-function even if it's not. This first
occurs in in fgetline(). So, I'll have to fix one of them to compile it.
Rod Pemberton
PS, I see you're having issues with seeing our posts, apparently Google
Groups related, if you eventually get to read this Steve, let me know...
s_dub...@yahoo.com
wrote:
> <s_dubrov...@yahoo.com> wrote in message
>
> news:e074e72c-456b-423e...@a11g2000yqm.googlegroups.com...> On Jun 24, 3:33 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > wrote:
> > > <s_dubrov...@yahoo.com> wrote in message
> > >news:bef0a6b9-1402-4b7b...@m9g2000yqe.googlegroups.com...
> > > > On Jun 22, 5:23 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > > > wrote:
> > > > > <s_dubrov...@yahoo.com> wrote in message
>
> >news:29b46bc6-66eb-488b...@28g2000yqu.googlegroups.com...
>
> ...
>
> > dosbox of XP won't load a .com > 64k, DOS will.
>
> Are you sure about that?
>
> IIRC, the limit is right around 64KB somewhere. Let me see if I can track
> that down before I post this ... I think I posted it a few years ago. Ok,
> I stated this some time ago:
>
> "DOS limits a .com to 0feffh (65280) bytes. That's 0ffffh (64KB) - 100h
> (256) for PSP. IIRC, MS-DOS won't load a .com even if it's size is close
> to 0feffh."
>
> Ed Beroset claims this changed around Win95 or due to it:http://groups.google.com/group/alt.lang.asm/msg/c9b50dad3761b942
>
> Dave Dunfield says MS-DOS 5.0 had the limit, actually 65279 or 65278, but
>
> > > > > On the win95 system I had to put the nasm commandline
> > > > (nasm.exe [blah]) into a .bat batch file to assemble it.
>
> > > Really? :-) Fortunately, Win95 was before my switch to x86 platform ...
>
> > Yes, fortunately for you..
>
> Well, actually, come to think of it, I don't recall exactly when I switched
> ... I used DOS prior to Win98, but I seem to be having a hard time
> remembering which year I got my first x86 machine. I think it was '93, but
> it could've been a few after that. I guess it just wan't an important
> event.
>
> Rod Pemberton
Sixth attempt to post, so briefly...
I did a self compile on a win98 system, it took nearly 11 minutes 8-/
Then again, if this post succeeds, well it took 3 days :-(
Steve
s_dub...@yahoo.com
wrote:
> "Rod Pemberton" <do_not_h...@noavailemail.cmm> wrote in message
>
> news:iueuta$v2a$1...@speranza.aioe.org...
>
> > <s_dubrov...@yahoo.com> wrote in message
> >news:58dd425b-1209-4e3e...@k16g2000yqm.googlegroups.com...
>
> > So, I didn't reply to your last post. Also, I haven't got around to
> > fixing the registers yet. I will.
>
> I changed the retrun registers in OSfn_() for AX and BX. So, instead of
> MOV BX,AX
> it's now:
> MOV BX, DX
>
> Hopefully, that's correct for AX and BX.
>
primary, that should work.
> However, I've run into another issue. I was ignoring an error parsing a
> comment, since it was "just" a comment... Apparently, that wasn't a good
> idea. It stopped parsing from there. After that was fixed, it seems the
> version of SmallC I've got cannot parse some of the C code in PCDOSLIB. It
> treats any return() as the end-of-function even if it's not. This first
> occurs in in fgetline(). So, I'll have to fix one of them to compile it.
>
Hmm, thanks for that info. That's the problem with adding more syntax
recognition, it gets used and the earlier compilers don't recognize
it. That why I leave earlier version available, for some traceability
on how to get here from what was.
> Rod Pemberton
> PS, I see you're having issues with seeing our posts, apparently Google
> Groups related, if you eventually get to read this Steve, let me know...
Yes, having issues, they all 'showed up' today, friday.
GG had a policy change on user profiles effective July 1, perhaps that
had something to do with the 4 day outage. I guess I'll have to
explore options for a plan B.
(fingers crossed)
Steve
Rod Pemberton
news:bbbc0706-e5aa-4bb7...@x12g2000yql.googlegroups.com...
> On Jun 29, 4:13 pm, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> wrote:
> > "Rod Pemberton" <do_not_h...@noavailemail.cmm> wrote in message
> > news:iueuta$v2a$1...@speranza.aioe.org...
> > > <s_dubrov...@yahoo.com> wrote in message
> >
>news:58dd425b-1209-4e3e...@k16g2000yqm.googlegroups.com...
> > However, I've run into another issue. I was ignoring an error parsing a
> > comment, since it was "just" a comment... Apparently, that wasn't a good
> > idea. It stopped parsing from there. After that was fixed, it seems the
> > version of SmallC I've got cannot parse some of the C code in PCDOSLIB.
It
> > treats any return() as the end-of-function even if it's not. This first
> > occurs in in fgetline(). So, I'll have to fix one of them to compile it.
> >
> Hmm, thanks for that info. That's the problem with adding more syntax
> recognition, it gets used and the earlier compilers don't recognize
> it. That why I leave earlier version available, for some traceability
> on how to get here from what was.
Was that something you fixed in one of your versions? Or, was it already
present?
> > PS, I see you're having issues with seeing our posts, apparently Google
> > Groups related, if you eventually get to read this Steve, let me know...
>
> Yes, having issues, they all 'showed up' today, friday.
On clax, Nathan found out it was a 2-day delay by GG for their new
interface.
> explore options for a plan B.
Unfortunately, Gougle Groups is the big service that allows one to read
Usenet via a webserver. Maxbaud had a way to read Usenet via web browser,
but I don't see that at the moment. It's possible there is a browser plugin
or widget for your browser somewhere. The real solution is to use a Usenet
news client: MS Outlook, Mozilla Thunderbird, connected to a good Usenet
server.
http://en.wikipedia.org/wiki/List_of_Usenet_newsreaders
Maxbaud has has resources for finding free or pay Usenet servers, groups,
etc.
http://freenews.maxbaud.net/
http://freeusenetnews.com/
For mailing lists (not Usenet), you can read via web browser using Gmane:
http://gmane.org/
I haven't tried these. These are supposedly usenet->web software:
UsenetWeb project
http://usenetweb.sourceforge.net/
OzWay
http://www.ozinsight.com/ozway/
If you have an NNTP client (MS Outlook, Mozilla Thunderbird, etc), you can
use some of the following. I've used AIOE and Cambrium. They're good.
AIOE's free servers, text-only, read and post Usenet news:
http://aioe.org/
news://nntp.aioe.org/
news://news.aioe.org/
Cambrium's free server, read-only, text-only Usenet news:
news://textnews.news.cambrium.nl/
There are other free, Usenet servers, but usually require account signup.
E.g.,
http://news.tornevall.net/
If you're really adventurous, you could set up a local webserver, use
news->html software, e.g., Ozway, UsenetWeb, then get a free feed, e.g.,
AIOE.
Rod Pemberton
s_dub...@yahoo.com
wrote:
> <s_dubrov...@yahoo.com> wrote in message
>
> news:bbbc0706-e5aa-4bb7...@x12g2000yql.googlegroups.com...> On Jun 29, 4:13 pm, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > wrote:
> > > "Rod Pemberton" <do_not_h...@noavailemail.cmm> wrote in message
> > >news:iueuta$v2a$1...@speranza.aioe.org...
> > > > <s_dubrov...@yahoo.com> wrote in message
>
> >news:58dd425b-1209-4e3e...@k16g2000yqm.googlegroups.com...
>
> ...
>
>
>
> > > However, I've run into another issue. I was ignoring an error parsing a
> > > comment, since it was "just" a comment... Apparently, that wasn't a good
> > > idea. It stopped parsing from there. After that was fixed, it seems the
> > > version of SmallC I've got cannot parse some of the C code in PCDOSLIB.
> It
> > > treats any return() as the end-of-function even if it's not. This first
> > > occurs in in fgetline(). So, I'll have to fix one of them to compile it.
>
> > Hmm, thanks for that info. That's the problem with adding more syntax
> > recognition, it gets used and the earlier compilers don't recognize
> > it. That why I leave earlier version available, for some traceability
> > on how to get here from what was.
>
> Was that something you fixed in one of your versions? Or, was it already
> present?
>
The original Cain small-c has trouble parsing comments across \n --
/*
** this form of comments causes problems.
*/
Does your version parse it?
-also 'do .. while' wasn't supported.
-also it didn't parse '&&' or '||', but did the '&' and '|' forms.
.. and probably a few more issues that I don't recall or was already
solved by the codeworks distribution.
CCNA changes how the '#include filename' is parsed, the name is now
copied to a static buffer and properly null terminated. Before, it
was parsed in the current line with \n termination which is a problem
when \n is CR,LF.
> > > PS, I see you're having issues with seeing our posts, apparently Google
> > > Groups related, if you eventually get to read this Steve, let me know...
>
> > Yes, having issues, they all 'showed up' today, friday.
>
> On clax, Nathan found out it was a 2-day delay by GG for their new
> interface.
>
More like 4 days for me.
> > explore options for a plan B.
>
> Unfortunately, Gougle Groups is the big service that allows one to read
> Usenet via a webserver. Maxbaud had a way to read Usenet via web browser,
> but I don't see that at the moment. It's possible there is a browser plugin
> or widget for your browser somewhere. The real solution is to use a Usenet
> news client: MS Outlook, Mozilla Thunderbird, connected to a good Usenet
> server.
>
Yes, however I never setup MS Outlook, and that saved me at least once
from maleware that tried to use Outlook in a back door fashion.
Yes, there are alternatives, thanks for your info, I'm still using GG
for the moment.
> http://en.wikipedia.org/wiki/List_of_Usenet_newsreaders
>
> Maxbaud has has resources for finding free or pay Usenet servers, groups,
> etc.http://freenews.maxbaud.net/http://freeusenetnews.com/
>
> For mailing lists (not Usenet), you can read via web browser using Gmane:http://gmane.org/
>
> I haven't tried these. These are supposedly usenet->web software:
>
> UsenetWeb projecthttp://usenetweb.sourceforge.net/
>
> OzWayhttp://www.ozinsight.com/ozway/
>
> If you have an NNTP client (MS Outlook, Mozilla Thunderbird, etc), you can
> use some of the following. I've used AIOE and Cambrium. They're good.
>
> AIOE's free servers, text-only, read and post Usenet news:http://aioe.org/
> news://nntp.aioe.org/
> news://news.aioe.org/
>
> Cambrium's free server, read-only, text-only Usenet news:
> news://textnews.news.cambrium.nl/
>
> There are other free, Usenet servers, but usually require account signup.
> E.g.,http://news.tornevall.net/
>
> If you're really adventurous, you could set up a local webserver, use
> news->html software, e.g., Ozway, UsenetWeb, then get a free feed, e.g.,
> AIOE.
>
Hmm, I think there is a learning curve to all that, we'll see.
Thxs,
Steve
> Rod Pemberton
Rod Pemberton
news:3c56f8b5-cece-4d81...@l18g2000yql.googlegroups.com...
> On Jul 2, 10:52 am, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> wrote:
> > <s_dubrov...@yahoo.com> wrote in message
> >
news:bbbc0706-e5aa-4bb7...@x12g2000yql.googlegroups.com...>
On Jun 29, 4:13 pm, "Rod Pemberton" <do_not_h...@noavailemail.cmm>
> > wrote:
> > > "Rod Pemberton" <do_not_h...@noavailemail.cmm> wrote in message
> > >news:iueuta$v2a$1...@speranza.aioe.org...
> > > > However, I've run into another issue. I was ignoring an error
parsing a
> > > > comment, since it was "just" a comment... Apparently, that wasn't a
good
> > > > idea. It stopped parsing from there. After that was fixed, it seems
the
> > > > version of SmallC I've got cannot parse some of the C code in
PCDOSLIB.
> > It
> > > > treats any return() as the end-of-function even if it's not. This
first
> > > > occurs in in fgetline(). So, I'll have to fix one of them to compile
it.
>
> > > Hmm, thanks for that info. That's the problem with adding more syntax
> > > recognition, it gets used and the earlier compilers don't recognize
> > > it. That why I leave earlier version available, for some traceability
> > > on how to get here from what was.
> >
> > Was that something you fixed in one of your versions? Or, was it already
> > present?
>
> [comments]
Uh, it seems you responded about comments... I asked if you fixed the
return() not being the end of procedure? Not important...
> The original Cain small-c has trouble parsing comments across \n --
>
> /*
> ** this form of comments causes problems.
> */
>
> Does your version parse it?
>
I don't believe so...
In C5LIB.C, the are all single line, usually:
/** stuff **/
In PCDOSLIB, I changed them to be that way too. It was probably because of
single line comments. I wouldn't have changed them otherwise... I do
recall that this wasn't handled in "my" version:
/** **/
That's a single space in there ... So, there is some bug in the "original"
still hanging around too.
> -also 'do .. while' wasn't supported.
> -also it didn't parse '&&' or '||', but did the '&' and '|' forms.
> .. and probably a few more issues that I don't recall or was already
> solved by the codeworks distribution.
Ok. I haven't gotten back to it.
> CCNA changes how the '#include filename' is parsed, the name is now
> copied to a static buffer and properly null terminated. Before, it
> was parsed in the current line with \n termination which is a problem
> when \n is CR,LF.
Discard CR ... If LF is \n and \r is CR, that'll work for LF and CR,LF. It
won't work for CR only (rare).
Rod Pemberton