The problems in comp.lang.c

[Friedrich Dominicus]

If you want to develop this idea, bear in mind the experience of some other
Usenet groups who use them, i.e. that Google Groups has been known to rip
tags out of subject lines posted via its servers. ("Do no evil"?
> Ha!)

I have no idea about that, but if that really happens I can not see
why one should stick to Google.

Regards
Friedrich

 

[Flash Gordon]

Anand Hariharan wrote, On 25/03/08 02:09:

Unless it is something very trivial (like misplaced closing parens),
I would be very interested in seeing an example of C code that
generated very different diagnostics from different compilers.

I was thinking of instances where one compiler produces a diagnostic and
another does not. Some compilers will diagnose certain "dodgy"
constructs even at there default levels, some will not diagnose some
"dodgy" constructs at any level.

Could you point me out to such a thread?

I'm sure I've seen one more recent than this, but here is one
http://groups.google.com/group/comp...?q=warning+gcc+msvc+group:comp.lang.c&lnk=ol&
I note on a quick scan one suggestion of asking somewhere compiler
specific, but that message did not say the original question was off
topic and the discussion about what warnings were appropriate continued
here without further complaint.

Could you clarify what you mean by /that/?

Licensing is not language specific and so not topical here. There are
other places where it would be more appropriate to discus how to put
your code under GPL.

Could I discuss how to package my code to facilitate a particular
license in c.l.c?

I would say not. We could discus methods by which you could embed a text
string in the binary (my favourite would be a command line option to
print out the license) but beyond that I cannot see how it could be
topical here.

I note that you are asking about topicality and not actually asking how
to do these things, and questions about topicality are always topical by
convention on Usenet.

 

[santosh]

No - the sort of thing that goes on is that various processes are
multi-tasked, without ever doing anything multi-tasking specific.
Or, for example, you ask for terminal input. The system doesn't
waste its time checking for keypresses. It goes off and does other
things until a keypress occurs, and then returns and presents its
result. No multitasking games required. But arranging for that
keypress to be reported on occurence is actually complex, and
probably involves various bufferes, interrupts, timers, process
priorities, whatever. It's in the OS, not the program.

However for many applications multithreading can dramatically improve
performance, since on most systems thread switching is much faster than
process switching. Another incraesingly important advantage of
multithreaded programs is that their threads can be scheduled
simultaneously under multiple CPUs or under multiple cores of a single
CPU. This _cannot_ be done with a program with single thread of
execution.

Consider a GUI program that compresses data and prints the progress in a
graphical window.

Now consider a multithreaded version that has two threads: one thread
that does the compression and another that redraws the window and
services GUI events. In this case the redrawing can proceed
simultaneously to the compression on truly parallel systems and
quasi-simultaneously on systems simulating parallel processing.

In the case of a single threaded version the code for compression must
voluntarily call the code for window drawing from time to time, while
event handling would become very complicated. A GUI event would
presumably send a signal to the program which a signal handler would
catch and set a global flag before returning. Then the event would be
serviced the next time the code calls service_pending_events().

As you can probably see multiple threads in a program makes life easier
in many types of programming tasks.

 

[santosh]

Anand Hariharan wrote:
[ ... ]

To be fair, c.l.c++ and a.c.l.l.c-c++ do tend to give more latitude
w.r.t topicality of posts.

[ ... ]

To a large extent, this has to do with C++'s language being orders
of magnitude more complex than C, and a lack of widespread
availability of implementations that conform to a standard that was
released more
than a decade ago. C90 has neither of these characteristics.

Actually I suspect that more C++ implementations conform more closely
to the 1998 C++ standard than C implementations do to the 1999 C
standard.

What I wrote (the part you have quoted above), I was referring to
C90, not C99. I have never paid serious attention to anything
introduced in C99.

Oops! A case of reading what I wanted to read.

 

[Richard Heathfield]

William Pursell said:

That is utterly absurd. Multi-threaded programs are popular,
but to claim that no one writes any single-threaded programs
is phenomenally naive.

strcmp("pointing out the absurdity of an article by",
"mounting a campaign against")
= 0; /* [1] */

Are you sure you can take the heat?

 

[Richard Heathfield]

Friedrich Dominicus said:

I have no idea about that, but if that really happens I can not see
why one should stick to Google.

Well, I can't see why now, either, but that doesn't change some people's
affinity for it. So let's follow an example:

(e-mail address removed) asks "compiler-specific question" (that's the OP subject
line, okay?).
(e-mail address removed) replies. In the process, he changes the subject line to
"[gcc] compiler-specific question".
(e-mail address removed) posts a second reply to the OP, but forgets to add the tag.
(e-mail address removed) posts a third reply to the OP. He adds the tag, but
Google strips it out again before posting it around the world.
(e-mail address removed) posts a reply to (e-mail address removed), in which the tag is retained.
(e-mail address removed) posts a reply to (e-mail address removed), never seeing the tag
(e-mail address removed) had added, and not thinking to add her own.
(e-mail address removed) posts a reply to (e-mail address removed), in which the tag is
stripped.

As you can see, even when people remember to add the tag in the first
place, if *any* Google people are active in a thread then the proportion
of correctly-tagged articles in that thread will tend to diminish. Asking
them not to use Google is like telling drivers of automatics to get a
stickshift (or Windows users to install Linux) - normally a complete waste
of time.

No, if you're going to tag, the fix is to find a tagging syntax that Google
doesn't strip. Some experimentation may be required.

 

[Antoninus Twink]

Malcolm McLean said:


It looks more like a lie to me. What did you expect from a troll? If he's
telling the truth, he'll be able to produce a message ID that proves his
claim, right? But he can't do that, because he isn't telling the truth.

I checked back in the Google Groups archive. Jacob's right that "Old
Wolf" was the one who first invented this particularly unpleasant slur,
but it was also repeated by "rosewater" (who shows all the signs of
being a sock puppet for Heathfield) at least twice:
http://groups.google.com/group/comp.lang.c/msg/240b64ceb789c798
http://groups.google.com/group/comp.lang.c/msg/4193730f7b8ae5d9

As Jacob said, it's not very pretty. Let's draw a veil over this grubby
corner of Heathfield's behavior.

 

[Paulo Jorge de O. C. de Matos]

Dear everyone

I've been lurking in this group for a few weeks now. Originally I had a
question to ask (about atomic/threadsafe operations in C), and I wanted
to get the feel of the newsgroup first. In fact I've got to know it well
enough that I realize there's no point in even asking my question, which
would be vilified as "off topic".

Hello all,

Let me contextualize my message. I've been actively asking questions
in clc for some years from 99 onwards. I've been away for some time
but now I've been following the group for the last 2/3 weeks. Yes, the
group might have problems, in fact probably more than it had before
but I don't see why people not replying to you about threads is a
problem at all. I do remember when I started posting to receive
replies such as "read the FAQ", or "that's off-topic", or "the C
standard knows nothing about displays". And you know what? I resorted
to unix programming, threads programming and such other newsgroups
which can actively help about those topics and learned the difference
about what C really is. In fact, I started resorting to K&R2 to know
if my C question might resemble some problem in the book before
posting.
"hummf, I have a question about threads... K&R2 says nothing about
threads. Better not post it in c.l.c. then."

From what I have observed, this group has big problems. There are some
positive posters here (I'd especially like to thank jacob navia for his
very interesting and useful long posts on stacks and debuggers), but
there seem to be many many more people here who post only to be
negative.

I don't know Jacob Navia but I would suggest the best path for Jacob
these days would be to have a blog. Posts on stacks and debuggers
would definitely be worth the reading through a RSS feed. I would then
use that knowledge to actively help those with their C
questions. There's no need, and no excuse to post questions unrelated
to the C language here, no matter how good they are. Saying this, let
me say I do think they Navia's articles are quite nice but would be
much better fitted into a blog. Not only due to the fact that posting
that here is off-topic but also because I could easily reach them
whenever I can, and even send the link to my g/f which is learning C
at the university and would definitely benefit from it.

The main example seems to be Richard Heathfield - I don't know what the
history of it is, but he obviously has a deep personal hatred of jacob
navia, and this dominates his posting. Most of his insults are pretty
puerile, but this constant negativity really seems to bring down the
atmosphere.

I need to say I am impressed to see you say that about Richard. I do
remember Richard Heathfield but in a slight different way. I remember
Richard Heathfield, Dann Corbit, Ben Pfaff, among others, actively
replying to my questions as a C student when I was actively posting
them. I might add that they were the reason why I was able to learn C
and enjoy C as much as I do these days. I remember to get a copy from
the US [when getting books from the US to Portugal was quite
expensive] as soon as "C Unleashed" came out and have loads of fun
while reading it. Damn, I still use it. If Richard did anything to the
community, it was surely not harmful.

I need to digress into talking about C99. I've been away these last
few years from USENET but I didn't expect that C99 had such a negative
impact. I use C mainly to implement SATisfiability solvers, which are very
processor intensive and every CPU cycle is indeed important. I use it
because is a very simple language, which I can easily tweak and where
I can understand what's happening deep down in it's translation to
machine code. Let me say I also use Scheme
regularly and I wouldn't dare to try to implement a state of the art
SAT solver in it because since it is too high-level leaves me
uncertain about the amount of machine code generated for a simple
expression. C is inherently a low level language and it should stay
that way. However, I regularly compile my code with -std=c99 [gcc option].
- 'Anywhere' variable declarations
for(int i = 0; i < 10; ++i) { ... }

are in my opinion extremely useful and make the code much more
readable (in fact, this is perhaps the only time I resort to them).
- // comments
help me document my code with simple line without the issue of
forgetting a missing */...

I guess these are in fact, the things I use the most from C99. The
other stuff might be useful to others. I even inquired the gcc people
why C99 is so poorly [in ~10 years they still miss a couple of
features] implemented. Their reply was ... interesting but
not as clear as one would wish:
http://gcc.gnu.org/ml/gcc-help/2008-03/msg00221.html

For those really pissed off with C99 and worried with the next one (is
there already any information about it or nobody knows what's coming?)
I would thing the best way out would be:
- Define a core C and an extended C (as someone already pointed out);
- Diverge from the standard C and define a de-facto standard like the
Scheme language did many years ago (IEEE has a published standard,
however, the community uses the RnRS, for n = 6 right now).

Anyway, I still think C is a VERY important language, for educational
and research purposes (not sure company-wise) and it should stay
on-track. I think adding heavy features like operator overloading (I
have heard someone talk about this recently)
might take the edge it has on other languages for being low-level. If
I wanted something high level I would go for C++ and Scheme (ordered
from the lowest to the highest level I can go).

I hope to keep track of the posts in c.l.c. and let's just enjoy
ourselves.

Have fun,

 

[Ben Bacarisse]

Jacob's right that "Old
Wolf" was the one who first invented this particularly unpleasant slur,
but it was also repeated by "rosewater"

Let's draw a veil over this grubby
corner of Heathfield's behavior.

I was going to add a comment, but it seems superfluous now.

 

[Richard Heathfield]

Ben Bacarisse said:

I was going to add a comment, but it seems superfluous now.

I see he's still at it. Having temporarily unplonked him (always, it seems,
a bad idea) to get the original context, I will simply point out that I do
not use sock-puppets, and that his claim is in error.

Such trolling is a constant drain on this newsgroup. If the people who do
it are genuine in their desire to improve the group, let them prove it by
posting intelligent, competent, and helpful answers to those who ask
questions about C here. And if they are not, why would anyone give their
slurs any weight at all?

 

[Kenny McCormack]

Ben Bacarisse said:


I see he's still at it. Having temporarily unplonked him (always, it seems,
a bad idea) to get the original context, I will simply point out that I do
not use sock-puppets, and that his claim is in error.

And I will simply point out once again that you are a lying sack of crap (*).

(*) Google it.

Such trolling is a constant drain on this newsgroup. If the people who do
it are genuine in their desire to improve the group, let them prove it by
posting intelligent, competent, and helpful answers to those who ask
questions about C here. And if they are not, why would anyone give their
slurs any weight at all?

Because the truth, although often painful, is always compelling.

 

[Ed Prochak]

jacob navia said:

Richard Heathfield wrote:
[snip]
Likewise, I assure you. But I think we differ over who we think the
morons are. If you don't like my articles, why not killfile me?
I would say the same. Good advice. You could use it yourself and
stop answering my posts.

Since it is in your commercial interest to have as few people as possible
pointing out your misunderstandings, blunders, and product plugs, your
response does not surprise me at all. It may surprise *you*, however, to
discover that I actually reply to relatively few of your articles. (It may
surprise others, too.)

This is the sort of thing I was talking about - a pointlessly negative
post. Every article of jacob's I've read has been well-informed and
useful - and if his posts put C in its wider context, instead of being
narrowly focussed on minor details, then so much the better!

In a discussion thread on topicality,
Navia calls Heathfield's views "stupid" and "bullshit"
Heathfield says Navia's posts contain "misunderstandings, blunders,
and product plugs"
and you say Heathfield's comment was "pointlessly negative"?

Morphius: "Do you think that is air you are breathing? Hmmm."

What I really can't understand is why people are /so/ exclusive. A
reasonable person would say that threads (which will certainly be part
of the next C++ standard, and I believe are also being considered for
the next C standard) are a topic of interest to C programmers, whereas
the breeding habits of gazelle aren't. And if something's borderline,
why not "be liberal in what you accept"?

You are the second person to suggest including discussions of
threading should be topical in c.l.c because it is being added to the C
++ language. Why would that be of interest to C programs?? C++ is a
different language! It has it's own group. By this argument, you might
be in favor of discussing some new feature of Transact-SQL in
comp.databases.oracle.server where is normally discussed PL/SQL
programming. Is that really your contention?

IMHO, Threads is not on topic in c.l.c, for example. Threads is a
perfect topic for comp.programming. Is it really so bad to send
discussions to the groups where they belong? That is why groups exist
by topic. These are not some AOL chat rooms. Try visiting
comp.programming There are some smart people there. Some of them are
even the same people as here! It is so much easier to get right
answers in the right groups.

Some groups are liberal in discussions. Some discussions just
naturally drift off topic on occasion. It can be beneficial to accept
borderline discussions, but not clearly off topic discussion. It is
beneficial to cut short threads that are borderline. It is fundamental
to the concept of groups: TOPIC is the key.

The reason it seems crazy to me is that it costs you (in Thunder Bird at
least) a single key-press to *ignore* a thread that you're not
interested in - you don't need to reply to it, or even take any time
seeing any more posts in the thread once you've ignored it. Making a
series of "that's off topic" posts takes /much/ more time than just
pressing the ignore key!

But if no one points out off topic posts, then they become more
frequent until the group loses its identity. There is no censorship in
unmoderated groups like this. Staying on-topic only happens when
people know what is on-topic. A certain amount of negative feedback
helps. Discussions like this are on occasion necessary in healthy
groups.

Ed

 

[Richard]

jacob navia said:

Richard Heathfield wrote:

[snip]

Likewise, I assure you. But I think we differ over who we think the
morons are. If you don't like my articles, why not killfile me?

I would say the same. Good advice. You could use it yourself and
stop answering my posts.

Since it is in your commercial interest to have as few people as possible
pointing out your misunderstandings, blunders, and product plugs, your
response does not surprise me at all. It may surprise *you*, however, to
discover that I actually reply to relatively few of your articles. (It may
surprise others, too.)

This is the sort of thing I was talking about - a pointlessly negative
post. Every article of jacob's I've read has been well-informed and
useful - and if his posts put C in its wider context, instead of being
narrowly focussed on minor details, then so much the better!

In a discussion thread on topicality,
Navia calls Heathfield's views "stupid" and "bullshit"
Heathfield says Navia's posts contain "misunderstandings, blunders,
and product plugs"
and you say Heathfield's comment was "pointlessly negative"?

Talk about cherry picking your examples.

Morphius: "Do you think that is air you are breathing? Hmmm."


You are the second person to suggest including discussions of
threading should be topical in c.l.c because it is being added to the
C

Lots of people have over time.

++ language. Why would that be of interest to C programs?? C++ is a

Because a hell of a lot of C programs feature multithreading. Linux and
Linux apps for one.

different language! It has it's own group. By this argument, you might
be in favor of discussing some new feature of Transact-SQL in
comp.databases.oracle.server where is normally discussed PL/SQL
programming. Is that really your contention?

IMHO, Threads is not on topic in c.l.c, for example. Threads is a
perfect topic for comp.programming. Is it really so bad to send
discussions to the groups where they belong? That is why groups exist

Why not discuss C threading in a C nesgroup? Is that so hard for you to
comprehend?

by topic. These are not some AOL chat rooms. Try visiting
comp.programming There are some smart people there. Some of them are
even the same people as here! It is so much easier to get right
answers in the right groups.

Agreed. Its a question of defining "right".

Some groups are liberal in discussions. Some discussions just
naturally drift off topic on occasion. It can be beneficial to accept
borderline discussions, but not clearly off topic discussion. It is
beneficial to cut short threads that are borderline. It is fundamental
to the concept of groups: TOPIC is the key.

And this post from you is?

But if no one points out off topic posts, then they become more
frequent until the group loses its identity. There is no censorship in
unmoderated groups like this. Staying on-topic only happens when
people know what is on-topic. A certain amount of negative feedback

Or when people simply say nothing when y question is not relevant to
their area of expertise. The problem here is that Heathfield wants to
limit it to HIS area of expertise.

helps. Discussions like this are on occasion necessary in healthy
groups.

Ed

There has been enough of a groundswell of public opinion here recently
which makes me think that topicality is indeed expanding in the C
related areas. And this can only be a good thing - C is used in the real
world and real world needs real solutions. Dont like the thread? Kill
it.

 

[Kenny McCormack]

....

There has been enough of a groundswell of public opinion here recently
which makes me think that topicality is indeed expanding in the C
related areas. And this can only be a good thing - C is used in the real
world and real world needs real solutions. Dont like the thread? Kill
it.

The primary justification given by the trolls (er, I mean, regulars) for
why they can't just "killfile it and be done with it" (thereby following
the general convention on the Usenet, as well as, I might add, the world
in general, of "If you don't like it, leave it alone") is that (they
claim) "The newbies will be confused".

This is just a variation on (and as valid as, which is to say, not valid
at all) society's overall obsession of "... for the chydlrun". It
sickens me.

 

[Ed Prochak]

The firm knowledge by most C programmers that most C appliations are
very similar to the ones they work on. I'm not sure why C programmers
are subject to such parochialism, but I've seen it over and over again:
Unix vs. MS-DOS, hosted vs. embedded, GUI vs. command line, etc. Each
side utterly convinced that their way is "obviously" dominant and unable
to comprehend how the other side could possibly disagree.

-Larry Jones

Pitiful. Just pitiful. -- Calvin

Still no data?

Okay so we know where that claim stands.

 

[Walter Roberson]

Because a hell of a lot of C programs feature multithreading. Linux and
Linux apps for one.

Linux has entire Usenet heirarchies to discuss its development.
Is there some reason why those heirarchies are not the appropriate
place for such discussions?

My "apps" work in Linux (no reason they shouldn't -- I wrote them
in standard C after all), but they do not use multithreading.

Linux is a unix-like operating system. Unix has been around for
more than 30 years, and many many Unix programs were written without
threads. Has someone discovered a new Church-Rosser Theorem, that
these days "threads" have to be added into the model of what
is computable or not? Or is what you are saying that programming
is no longer about what is being computed and is instead about
the "user experience", and all those programs that used to
-compute- something are just living on life support?

 

[Keith Thompson]

Hm, I mentioned mailing lists, but I can not see why the should not be
of "broader" interest and this group would fit for that IMHO.

You snipped the context of my remark. The subject was libapr,
http://apr.apache.org/.

I don't believe it would be possible to impose such a convention in an
unmoderated group.

Hm, the OP uses [someTag] on reply on usually gets the subject for
free so the reply would also be tagged if not changed willingly by the
replier. It's very non-obtrusive IMHO.

Sure, if the OP *always* uses the right tag. Many threads here are
started by newbies asking questions (nothing wrong with that, of
course). We can't even get newbies to read the FAQ before posting; do
you think we can get *everyone* to follow a new tagging convention?
If an original article doesn't have the correct tag, it's impossible
to add it consistently. And changing subject headers within a thread
makes it difficult to keep track of a discussion, even assuming that
newsreaders manage threads by References rather than by Subject.

Such a convention could work well in a moderated newsgroup (but
comp.lang.c.moderated has such low volume that it's not necessary
there).

Again, you snipped so much context that this doesn't make much sense.
(I was talking about the appropriate place to discuss
multi-threading.)

standardization suggestions? should then be taken to comp.c.xxx?

and not in the broader group here?

Yes, suggestions for new features are appropriate in comp.std.c,
though personally I don't mind when existing discussions here veer
into that area. (And if comp.std.c didn't exist, such discussions
would certainly be topical here.)

 

[Ed Prochak]

Great.

But... did you know?

o There are quite a few parallel extensions to C.
o There are several packages that implement threads in C, together
with special syntax.

o Thraeds or not thraeds? Wouldn't arrays, a naturally parallel data
structure be better than making threads and threads? Is the thread
model scalable?

Beyond your straw man, we could discuss *issues* that pertain to
language development. Because if you can't implement it in your
language of choice, many models just do not get used. There were a lot
of "parallel languages" around, and still there are. But they do not
catch into the mainstream.

That could mean that we could see how parallel concepts fit into the
language.

Now get out of your straw man and let's analyze the real issues.

This group is becoming a BORE. Why?

Because by narrowly focusing in questions like:

ON TOPIC QUESTION
------------------------
Hi guys

I wrote
i = i++ - --i;

and it doesn't work...
------------------------

Why can't we discuss the things that are related to C but not in
the standard?

Just try to answer without straw mans, and getting into the arguments of
the other side

Thanks

I would think discussions of adding parallelism features would go to
comp.std.c If you are using a C compiler that has such features
already, those extensions should be discussed in the group/mailing
list/website dedicated to that compiler. I see no justification to
bring those topics here.

Threads are already in their own group comp.programming.threads
There are groups for networking, databases, embedded, and real-time
programming.
Some of the questions being proposed for this group are well covered
in those groups. Others clearly could start in comp.programming and if
they came across a C specific question could then post that question
here.

Ed

 

[Ed Prochak]

I believe there are mailing lists that discuss it. Not everything
must have a newsgroup where it can be discussed.

Hm, I mentioned mailing lists, but I can not see why the should not be
of "broader" interest and this group would fit for that IMHO.

I don't believe it would be possible to impose such a convention in an
unmoderated group.

Hm, the OP uses [someTag] on reply on usually gets the subject for
free so the reply would also be tagged if not changed willingly by the
replier. It's very non-obtrusive IMHO.

It is not the replies that are the issue. All the Tag does is move the
question of topicality from which newsgroup to which Tag. IOW it does
nothing to solve the problem of off topic posts.

standardization suggestions? should then be taken to comp.c.xxx?

comp.std.c for C

and not in the broader group here?

No.


Ed.

 

[Kelsey Bjarnason]

[snips]

o There are quite a few parallel extensions to C. o There are several
packages that implement threads in C, together
with special syntax.

And as long as they're doable in standard C, bring 'em on. If not, well,
I *could* create an extension to C which allows VB syntax, but I don't
think anyone other than you would regard VB as topical here.

Beyond your straw man, we could discuss *issues* that pertain to
language development.

Oh? What issues would those be? Things such as "how do I do _this_ in
C?" Of course those are topical - as long as they're things actually
doable in C. See how easy this is?

This group is becoming a BORE. Why?

Because boneheads like you refuse to grasp the simple concept that it is
about C, not somebody's screwed up VB import modules, and thus persist in
bringing up unwanted, non-topical, irrelevant crap day in and day out
meaning that others need to keep trying to educate you that the group is
about C. Unless and until you folks learn this lesson, the same issues
will be endlessly repeated, making it boring.

Why can't we discuss the things that are related to C but not in the
standard?

Well, VB is related to C: they're both programming languages, after all.
So let's talk VB. Sounds good to me. Let's open the door to *all*
programming languages.

No, thanks, I'd rather stick to C. If you want those other languages,
there are other groups for them.