help me learn C

[Chris Hills]

Surely there's a more appropriate group in which to discuss AIDS than
the embedded newsgroup.

Yes but it came up as a side comment in a topic that strayed a bit. The
interesting comment from most was that it did no harm. Also that
permitted the NG members to have a "social" off topic chat that was felt
to be a good idea every now and again. A bit like a tea/coffee break
when you discuss football or the latest government **** up.

It turns out that the "vociferous few" such as Brian are also the most
uniformly knowledgeable and helpful posters, so if you'd like to
migrate to their killfiles, by all means keep belaboring this point.

OK... as I am on various other NG's including the ISO WG's it should not
be to much of a problem.

You are clearly new here, as our most cranky regular, Dan Pop, has not
been seen in some time.

He is alive and well on other lists and as interesting as usual. I
would not presume to call him cranky.

 

[Chris Hills]

As we all know, there is no written charter simply because the
newsgroup was created before the existence of newsgroup charters.
As a result, topicality is judged by the consensus of the regulars.
I've already stated what I think that consensus is and should be.

OK.

Can you explain, with some *specific* examples, what kinds of things
you think should be topical here that people have said shouldn't be?

Not *specifically* That is where you and I differ. I do not want a
rigid definition of what is on topic. Rigid definitions are OK for
system specifications and standards* but not for a discussion area. You
need broad definitions. Anything else stifles debate.

*In which the current C standard falls over badly.

In a recent [according to you] off topic post the reply was an answer to
some parts but the OP was directed to another NG for other parts
(hardware) of the question. He was also told which parts of his code
were not standard C and should be amended... another void main()!

This is a much better response than just saying "go away". Because they
do and start up other NG's and lists. Many others find those lists
before they find this one and you loos a lot of people who would be
useful contributors here. Actually there has been some discussion on the
general decline for usenet anyway.

I do share your fear that this could end up as an MS-C/ windows NG but
given the number of Linux and embedded users there are this should not
be the case. Though few of them use standard C. Also this is your
opportunity to educate.

(BTW do you include IEC, MISRA and ECMA as part of the standardisation
process or just ISO?)

In a communication today regarding MS compilers I was told the
following:-

now in the works, plus the ECMA standards cited above, all describe
*working products that have been shipped to millions of customers.*
By contrast, both the C99 and C++ Standards have been widely, and
properly, criticized for codifying nonexistent practice which explains
why 100 per cent conformance to either standard is rare.

So the C99 standard has "codified nonexistent practice" and explains why
with I think 4 exceptions no one has implemented a full C99 compiler. I
believe that 2 of the 4 were implemented for political reasons and will
probably not be used in full conforming mode.

So C is neither implemented or used as per the standard and since you do
not discuss the standard here that is on comp.std.c and you do not
discuss the language as implemented discussion will tend to zero over
time. Or be pointless.

All I want to do is let people ask general C questions here and have
sensible conversations.

(By convention, meta-discussions about topicality, like this one, are
considered topical.)

In my case I don't need any dispensation as I don't subscribe to your
rules in the first place.

 

[Malcolm]

Given that some of use have over 25 years professional experience some
one at age 14 is not going to have any real world experience.

He is not old enough to have had any work experience nor completed any
formal training. Two months no this NG is hardley enough either.

Adults minds are better than children's, but only the best of them.

I was happily programming video games at 14. Now I know a lot more than I
did then, but the fundamentals haven't changed. 14 year olds' programs still
have to execute correctly and in reasonable time, just like professionals'.
Of course the consequences of error may be less serious. Or maybe more
serious - a really good program by Andrew might have much more impact on his
future earning prospects than a really good program by someone in a senior
position.

I wouldn't put too much store by either work experience or formal training.
Formal training is a bit like "creative writing" courses. There are
individuals who are helped, but basically if you have it in you to be a good
programmer or a good novelist then you can achieve that by practise and
hanging around this ng. Work experience often doesn't translate well to
other workplaces. For instance I once applied for a job in medical imaging
software. The interviwer asked me what formal methods I used. I replied that
we didn't use formal methods in games programming. She didn't seem to
believe me and said that surely you have a method for going from idea to
finished product. I said we didn't use any formal method, and of course I
didn't get the job.

I only issue I've got with Andrew's age is some of the language used by
other posters. For instance jacob navia was described in another thread as a
"fuckwit". He probably is, but it isn't fair to minors for them to come
across these words in a technical forum.

You have an opportunity to build a much larger community on here. You
can educate people in the differences between what they are doing and
standard C. Make them aware there is a difference. In some cases there
may be a standard or portable way of doing things.

I would like to read every post every day. I cannot achieve that. The
comp.lang.c community would in fact fall apart if it got much larger.My website www.personal.leeds.ac.uk/bgy1mm
Programming goodies

 

[Malcolm]

As a newsgroups traffic expands, it becomes unwieldy and
participants drop out. This puts a natural cap on the size of any
one group. It is something like the relationship of the population
of rabbits to the population of lynxes.

No its not.
Read about optimal group size theory in "Living in groups" by Jens Krause.

The idea is that each individual can choose whether to join or leave a group
based on his selfish considerations. Normally there will be certain
advantages to membership, such as protection from predators, or, in our
case, exchange of information. However there will also be penalties, such as
competition for food, or, in our case, time invested in reading through the
ng. There are also costs in joining and leaving groups, or finding new ones.

For instance, there might a forum on yahoo groups which meets my needs much
better than comp.lang.c, but I've got to find it and then establish myself
as a regular, which all takes time and effort.

So the size of the group is determined by these complex factors, and may be
bigger or larger than what would be regarded as optimal.

Website: www.personal.leeds.ac.uk/bgy1mm
Programming goodies.

 

[Keith Thompson]

OK.


Not *specifically* That is where you and I differ. I do not want a
rigid definition of what is on topic. Rigid definitions are OK for
system specifications and standards* but not for a discussion area. You
need broad definitions. Anything else stifles debate.

But surely you can provide specific examples of things that you think
should be considered topical here.

*In which the current C standard falls over badly.

In a recent [according to you] off topic post the reply was an answer to
some parts but the OP was directed to another NG for other parts
(hardware) of the question. He was also told which parts of his code
were not standard C and should be amended... another void main()!

Without knowing the example you're referring to, I can't really say
much about this. If, as you say, the poster was told that parts of
his code were non-standard and should be corrected, and that other
parts of his code were system-specific and should be discussed in
another newsgroup, it sounds like the query was dealt with properly
and in accordance with what I think is the consensus among the
regulars of what is and is not topical.

If you'll cite a specific thread, perhaps we can discuss this further.
Based on this vaguely described example, I don't see where we
disagree, and yet it seems clear that we do.

This is a much better response than just saying "go away". Because they
do and start up other NG's and lists. Many others find those lists
before they find this one and you loos a lot of people who would be
useful contributors here. Actually there has been some discussion on the
general decline for usenet anyway.

I do not advocate simply telling people to "go away" with no further
explanation (except for deliberate trolls, of course). (No, I don't
consider you to be a troll.)

I do share your fear that this could end up as an MS-C/ windows NG but
given the number of Linux and embedded users there are this should not
be the case. Though few of them use standard C. Also this is your
opportunity to educate.

(BTW do you include IEC, MISRA and ECMA as part of the standardisation
process or just ISO?)

IEC is part of ISO, isn't it, or at least closely associated with it?
The C99 standard is "ISO/IEC 9899:1999 (E)". So yes, certainly IEC is
part of the standardization process. I'm less sure about MISRA; I
tend to think of it as a secondary standard like POSIX, but I'm not
sure where it should be discussed. I'm not aware of any ECMA standard
for C.

[...]

So the C99 standard has "codified nonexistent practice" and explains why
with I think 4 exceptions no one has implemented a full C99 compiler. I
believe that 2 of the 4 were implemented for political reasons and will
probably not be used in full conforming mode.

So C is neither implemented or used as per the standard and since you do
not discuss the standard here that is on comp.std.c and you do not
discuss the language as implemented discussion will tend to zero over
time. Or be pointless.

We frequently have discussions about the differences between C90 and
C99, and the fact that C99 as isn't widely implemented as some of us
might like. When I cite a standard, I usually quote from C99 (usually
N1124), largely because my copy of the C90 standard was rendered into
PDF in such a way that it's very difficult to search or to
copy-and-paste.

All I want to do is let people ask general C questions here and have
sensible conversations.

I agree 100%. We just need to figure out what the phrase "general C
questions" means. I have yet to get a good sense of what you mean by
that phrase.

 

[Richard Heathfield]

Chris Hills said:

Not *specifically* That is where you and I differ. I do not want a
rigid definition of what is on topic. Rigid definitions are OK for
system specifications and standards* but not for a discussion area. You
need broad definitions. Anything else stifles debate.

So anyway, Chris - how's the garden?

I do share your fear that this could end up as an MS-C/ windows NG but
given the number of Linux and embedded users there are this should not
be the case.

Sure. Instead, it'll end up as a complete mess, with large quantities of
Linux articles (for which comp.os.linux.development.apps already exists),
large quantities of Windows articles (for which
comp.os.ms-windows.programmer.win32 already exists), and about the same
number of questions about the core language as we get now, except that
they'll be harder to find, and some of us are too busy to spend time
weeding out the junk. So you've come up with a great way to purge this
group of C expertise. Nice one.

Though few of them use standard C.

Those of us who /do/ use standard C would like a newsgroup in which we can
discuss it. And oh look, we /have/ such a group already. If people want to
discuss libs or platforms, there are already lots of newsgroups for doing
that.

So C is neither implemented or used as per the standard

I agree that C99 isn't implemented sufficiently widely to be useful. In
fact, the only impact C99 has had on me is to restrict me to the common
subset of C90 and C99, against a time when C99 /might/ one day be widely
implemented.

But C90 is still a viable standard, and whilst I accept that there might be
esoteric little corners of the language that some compilers don't get quite
right, I keep my code in the mainstream. (For example, I don't care about
the rights and wrongs of p = p->next = q; when I know I can write p->next =
q; p = p->next; and sidestep the issue completely.)

Because C90 is a viable standard, and because I write so much code in C90, I
find it very useful to subscribe to a newsgroup where many C experts are
ready to discuss and advise on standard C issues. This is the only
newsgroup I know of that provides such a service. If it goes down the
tubes, I will not be best pleased. And if you open it up to discussions
about Unix, Linux, the Mac, Windows, windows, window panes, greenhouses,
tomatoes, gardening, garden tools, spades, hearts, diamonds, emeralds, the
Irish question, Parnell, Gladstone, Gladstone bags, carpet bags, vacuum
cleaners, vacuums, vacuum fluctuations, big bang, little bang, internal
combustion, the price of oil, and the problems inherent in invading enough
countries to keep a tight control on the supply, then the comp.lang.c group
will indeed go down the tubes, and rebuilding it afterwards will not be at
all easy.

All I want to do is let people ask general C questions here and have
sensible conversations.

Right. So we want the same thing, yes? Except, of course, that we don't. I
don't consider a question about, say, getch() or socket() to be a general C
question.