Downloadable standards file?

[matevzb]

But do they claim *conformance* to C99? If not, you don't know whether those
compilers will compile your C99 programs.

Some of them do, to a certain extent of course. But then there's
usually some kind of smallprint (e.g. untested, this doesn't compile
yet etc.) so it's generally very time consuming to figure out what one
can use on all platforms.

In general I guess Jacob could be semi-right regarding issues on
portability, so maybe there should be say comp.lang.c90,
comp.lang.c.portable or even comp.unix.posix_in_real_life =)
I'll take my portability issues elsewhere next time...

 

[Richard Heathfield]

jacob navia said:

Richard Heathfield a écrit :

Look, IBM says explicitely that they conform to the standard.

Compiler, yes. No libc.

Comeau compiler
under windows says explicitely it conforms.

I found no such statement. Cite?

Of course you want a new standard that is implemented 100% everywhere.

I'd want at least two of Borland, Microsoft and gcc to provide a conforming
compiler, together with at least one mainframe compiler (LE370 or similar)
and one Mac compiler.

That can't be, so you will always find something.

The situation is much worse in C++, where there was in a recent
survey by Plauger almost no compiler fully conforms.

Would you say then that programming in C++ is impossible?

No, you can write programs in C++ to your heart's content - you just can't
compile them (unless, of course, you are using the compilers that do
conform).

<snip>

 

[Richard Heathfield]

matevzb said:

In general I guess Jacob could be semi-right

If only. It would be a 50% improvement on his usual performance.

regarding issues on portability,

Mr Navia doesn't understand portability, as his track record here in clc
shows clearly.

 

[Keith Thompson]

Why should this group be obsessed with writing portable
programs? This is not the subject of this group but
the C language and the C standard as it is.

You do not like it?

Make a new group:

comp.lang.portable

where you discuss portability to all toasters of the world
where they run in the famous embedded processor that
doesn't use the bss -) or any other stuff for that
matter.

C99 is currently the official standard *according to ISO rules*.

You seem to be asserting this newsgroup is required to follow ISO
rules. Given the fact that the C90 standard is still significant, I
can see no basis for such a requirement.

There is no formal charter for this newsgroup, but the general
consensus seems to be that C99, C90, and even K&R and earlier versions
of C are considered topical. If you don't like *that*, make your own
newsgroup.

 

[CBFalconer]

matevzb a écrit :

Excuse me but why I should care where YOUR code runs on?
If you do not want to use C99 it is your decision. I accept
that but the current standard is not C90, and as such there is
no point in trying to move backwards all the language because
of a possible problem with your code.

Which platform? What problems did you encounter?
You do not specify...

Why should he? He isn't writing for HIS compiler or platform. He
is writing to not care what compiler is used. He has high
confidence in the results if he avoids C99 specific abilities.

 

[CBFalconer]

.... snip ...

<OT>It should run on HPUX, Windows, Solaris, Linux, Tru64, IRIX,
Sinix, AIX, Netware, SCO (OpenServer and UnixWare) to name the less
problematic ones. Then there are MPE/iX and OpenVMS (maybe AS/400 as
well). Note that gcc is only used for Linux (not my decision) and
that support for multiple versions of each system is required.</OT>

I believe gcc runs on at least HPUX, Windows, Solaris, Linux, AIX.
Quite likely some, if not all, of the other named ones.

Here are a few of the systems it runs on, according to "info gcc":

_Machine Dependent Options_
*Note Hardware Models and Configurations: Submodel Options.

_M680x0 Options_
_M68hc1x Options_
_VAX Options_
_SPARC Options_
_Convex Options_
_AMD29K Options_
_ARM Options_
_MN10200 Options_
_MN10300 Options_
_M32R/D Options_
_M88K Options_
_RS/6000 and PowerPC Options_
_RT Options_
_MIPS Options_
_i386 and x86-64 Options_
_HPPA Options_
_Intel 960 Options_
_DEC Alpha Options_
_DEC Alpha/VMS Options_
_Clipper Options_
_H8/300 Options_
_SH Options_
_System V Options_
_ARC Options_
_TMS320C3x/C4x Options_
_V850 Options_
_NS32K Options_
_AVR Options_
_MCore Options_
_MMIX Options_
_IA-64 Options_
_D30V Options_
_S/390 and zSeries Options_
_CRIS Options_
_PDP-11 Options_
_Xstormy16 Options_
_Xtensa Options_

 

[CBFalconer]

.... snip ...

Would you say then that programming in C++ is impossible?

Or nearly so.

 

[jacob navia]

Richard Heathfield a écrit :

matevzb said:




If only. It would be a 50% improvement on his usual performance.




Mr Navia doesn't understand portability, as his track record here in clc
shows clearly.

Translation:

"Mr Navia doesn't agree with me. Since I can't accept that fact
(who would DARE to disagree with me?) it is just that he doesn't
UNDERSTAND portability"



jacob

 

[Richard Tobin]

C99 is currently the official standard *according to ISO rules*.

And C90 is no longer an ISO standard.

Since it's clearly still a standard, but not a de jure standard,
it must be a de facto standard. Like, say, gcc...

-- Richard

 

[Keith Thompson]

And C90 is no longer an ISO standard.

Since it's clearly still a standard, but not a de jure standard,
it must be a de facto standard. Like, say, gcc...

gcc has its own set of newsgroups.

There are several ways the topicality of this newsgroup *could* be
defined. The way it should be and is defined, by a fairly broad
consensus, includes C90 but excludes gcc-specific extensions.

 

[Richard Heathfield]

jacob navia said:

Translation:

"Mr Navia doesn't agree with me. Since I can't accept that fact

Sure I can.

(who would DARE to disagree with me?)

Lots of people. They do it all the time. And sometimes they're right and I'm
wrong. When that is the case, I generally say so.

it is just that he doesn't UNDERSTAND portability"

Check Mr Navia's track record in this group, and it will become evident that
this is indeed the case.

 

[Randy Howard]

Richard Heathfield wrote

jacob navia said:



No reason at all. That's the whole point of portability. It simply doesn't
matter where his code runs.


Quite so. Even if he had a C99-conforming compiler, which is doubtful, it
appears that he'd rather use something that stands a chance of porting to
other platforms.


It's true that C90 is not the current de jure standard, but it remains the
current de facto standard for those who require portability.

You left out "... or for those that require a working
implementation conforming to a published standard for their
target platform(s)".

Fine, so let's move the language forwards by getting some C99-conforming
implementations out there. There are way too few right now to make it a
viable standard.

Well, given that it'll soon be a decade later, and still people
will be searching for "C99" compilers (or have just given up by
then), what's the point? The incredibly small number of people
that actually have a deep an abiding need for a C99 compiler
must have found one by now, or written their own. The rest of
the C programming universe obviously doesn't care enough to
solve the problem. Some problems just don't need solving.

 

[Randy Howard]

jacob navia wrote

Look, IBM says explicitely that they conform to the standard. Comeau
compiler under windows says explicitely it conforms.

Yippee. Yawn.

Of course you want a new standard that is implemented 100% everywhere.

Who said they wanted that? Hint: C99 isn't a "new standard".
Furthermore, it's nowhere /near/ being implemented 100%
everywhere. Laughable.

That can't be, so you will always find something.

What decade will it be available for 90% of the platforms out
there? 100 years after that answer is positive for C90
compilers?

The situation is much worse in C++, where there was in a recent
survey by Plauger almost no compiler fully conforms.

Almost no C++ programmer fully conforms to C++ either.

Would you say then that programming in C++ is impossible?

Almost a "Yes", regardless of the survey data. *cough*

And if some compiler has a bug, or a non-implemented stuff, so
what?

Spoken like a true non-standard compiler developer.

(What is even more important than standards conformance)

Having a compiler that is even pretending to implement the
standard you wish to have conformance with in the first place
would be a good start.

 

[Randy Howard]

Richard Heathfield wrote

jacob navia said:

Compiler, yes. No libc.

[snip]

I'd want at least two of Borland, Microsoft and gcc to provide a conforming
compiler, together with at least one mainframe compiler (LE370 or similar)
and one Mac compiler.

You partially in luck, because gcc runs on Mac OS X. So,
if/when gcc gets there for Windows/dos/whatever, it'll get there
for the Mac. You also seem to be leaving out probably the
biggest set of C developers still in the wild, those writing
code for embedded environments, but of course most of them
probably have no use for c99 at all. Then again, neither do
most of the rest.

Of course, as you mentioned above, you also need a libc to go
along with it.

 

[jacob navia]

Randy Howard a écrit :

jacob navia wrote



Yippee. Yawn.




Who said they wanted that? Hint: C99 isn't a "new standard".
Furthermore, it's nowhere /near/ being implemented 100%
everywhere. Laughable.




What decade will it be available for 90% of the platforms out
there? 100 years after that answer is positive for C90
compilers?




Almost no C++ programmer fully conforms to C++ either.




Almost a "Yes", regardless of the survey data. *cough*




Spoken like a true non-standard compiler developer.




Having a compiler that is even pretending to implement the
standard you wish to have conformance with in the first place
would be a good start.

Who are you to give me lessons man?

Do *I* tell you what you have to do?

I have spend the last 5 years working in a C99 implementation
that I distribute at no charge. As you noticed elsewhere,
a full libc implementation is also needed, not to mention the
code generation part.

I am nowhere near 100% complicance but I have implemented most of this
standard. And I have done it without any big companies behind
me like gcc and Red-hat/IBM, that contribute to their quite big budget.

A completely new implementation of printf/the math library/ many strings
functions, etc etc.

Plese keep this in mind:

I am not part of any standards comitee, and I do not earn a penny more
if you use the current standard or you come back to 1989 or even
earlier...

Why not K&R C?

There are many embedded compilers producing code for yesteryear's
coffee machine that are still in K&R C, so use THAT...

 

[Randy Howard]

jacob navia wrote

Who are you to give me lessons man?

I'm not trying to give you lessons. You asked a question, and I
answered it with my opinion.

You have historically shown a distinctively low absorption rate
for accurate information delivered from others anyway, so I
won't bother hoping for you to learn anything from me, or anyone
else for that matter.

As to the "who are you", I'll just say someone that has been
writing C programs, portable ones, on a whole lot more
compilers, platforms and CPU architectures than you've ever even
been in the same room with, since you were in diapers, if I had
my guess. Enough of that...

Do *I* tell you what you have to do?

No. But it's a stupid question, since I haven't told you what
you have to do. More importantly, I haven't told anyone here
what they have to do.

I have spend [sic] the last 5 years working in a C99 implementation
that I distribute at no charge. As you noticed elsewhere,
a full libc implementation is also needed, not to mention the
code generation part.

Good for you.

I am nowhere near 100% complicance [sic] but I have implemented most of this
standard.

So somewhere between "most" and "nowhere near 100%", whatever
that means.

And I have done it without any big companies behind me like gcc
and Red-hat/IBM, that contribute to their quite big budget.

I didn't realize gcc was a big company. I thought it was a C
compiler, and a pretty damn good one, particularly when invoked
with the correct command line arguments so that it doesn't
compile a variant language which I don't want to use.

Plese keep this in mind:

I am not part of any standards comitee [sic],and I do not earn a penny more
if you use the current standard or you come back to 1989 or even
earlier...

I don't give a popcorn fart whether you make any money or not.
That wasn't even remotely related to what you replied to. I
have no idea why it's even an issue.

Why not K&R C?

I have written a lot of K&R C in the past. Today, every
platform that I care about has a workable C89/90 compiler that
will compile code and generate the expected code for same.

Furthermore, that C standard provided a lot of very useful
features not available with K&R C, so it is worth taking
advantage of them.

If I had a reason to work on a platform for which no C89
implementation were available, I'm sure good old K&R C could fit
the need.

I can't even pretend to claim that all those platforms have C99
support. Not even a tiny subset of them. If otherwise, I might
feel differently, although the vast majority of the C99
extensions I wouldn't ever use even if ubiquitously available.

/NONE/ of the extensions you have put into your "navia-C"
non-standard compiler would ever be used in any of my code,
because they are present not only in a single compiler, but for
a single processor family and single operating system (minus a
few MS branding changes and service packs). In other words,
100% worthless for someone interested in portable code
development.

There are many embedded compilers producing code for yesteryear's
coffee machine that are still in K&R C, so use THAT...

So, I never told you what to do, but you pretended to be
offended that I did, and here you go, doing the same. Hypocrisy
noted.

Well guess what, I have quite a few compilers handy here for
various embedded controllers, and all of those I am interested
in have C89/90 support. NONE of them even pretend to be C99
compilers. In the embedded space, there is nothing even
remotely interesting about C99.

 

[jacob navia]

Randy Howard a écrit :

/NONE/ of the extensions you have put into your "navia-C"
non-standard compiler would ever be used in any of my code,
because they are present not only in a single compiler, but for
a single processor family and single operating system (minus a
few MS branding changes and service packs). In other words,
100% worthless for someone interested in portable code
development.

The version of lcc I am working with runs in Power-PC, linux,
windows, and several embedded processors. You just have no idea
and you just speak out your prejudices.

But surely my compiler is not for you anyway. Keep
away from it, and put me into your kill file.

OK?

Thanks in advance for uour understanding.

 

[Randy Howard]

jacob navia wrote

Randy Howard a écrit :

The version of lcc I am working with runs in Power-PC, linux,
windows, and several embedded processors. You just have no idea
and you just speak out your prejudices.

Do I need to list all the combinations that are important for
portable programming that your list does not include? No, what
would be the point?

But surely my compiler is not for you anyway.

Yes, I've tried it. I'm well aware of this fact.

Keep away from it, and put me into your kill file.

Who's telling who what to do again? Your repeated hypocrisy is
again noted.

 

[Mark McIntyre]

So What?

Is this group called:
comp.std.portable
???

Yet again, you display not only your contempt for the regulars here,
but also your incredible arrogance. I despair of you, its a real shame
to see someone so evidently intelligent throw away so much because of
a petty dispute.
--
Mark McIntyre

"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it."
--Brian Kernighan

 

[Mark McIntyre]

And C90 is no longer an ISO standard.

Since it's clearly still a standard, but not a de jure standard,
it must be a de facto standard.
mhm.

Like, say, gcc...

False extrapolation. gcc is an implementation of a standardised
language.
--
Mark McIntyre

"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it."
--Brian Kernighan