Open letter to Mr Thompson

[Default User]

I have to ask, yet again, why people keep feeding this troll. Is there
really value added to the newsgroup with these constant flamewars with
Mr. Navia?




Brian

 

[jacob navia]

Seebs a écrit :

I haven't seen the proposal. However, there are many proposals I've seen
over the years which were such that there was nothing to collaborate ON;
it was just plain a bad idea.

Sure but I wonder if you read this newsgroup...

I have started several threads here and in comp.std.c about fixing
the overflow problem

c = a+b;

if the operation overflows you are not warned and there is no
way to know. Horrible isn't it?

I proposed

#pragma STDC OVERFLOW_CHECK on-off-flag

See the thread "Zero overhead overflow checking"

Note that I wrote now

on-off-flag

and not

on_off_flag

as I had originally because thompson told me it is better with - and
not with _. But beyond that, the first substantive proposal he did was
that there is ALSO an overflow when you say:
char c = integer;

and integer is > CHAR_MAX

That is why I proposed him to do a parallel proposal.

Conclusion:
he treats me of a jerk.

Go figure

 

[Seebs]

Sure but I wonder if you read this newsgroup...

Not often.

I have started several threads here and in comp.std.c about fixing
the overflow problem

c = a+b;

if the operation overflows you are not warned and there is no
way to know. Horrible isn't it?

Times this has ever harmed me:

Maybe two.

Times I've benefitted from code which ran faster because it didn't check
this:

Maybe thirty million.

as I had originally because thompson told me it is better with - and
not with _. But beyond that, the first substantive proposal he did was
that there is ALSO an overflow when you say:
char c = integer;

and integer is > CHAR_MAX

That is why I proposed him to do a parallel proposal.

I don't see either of these proposals as adding real value.

Conclusion:
he treats me of a jerk.

So, basically, he told you he didn't think your idea was a good idea,
and you responded by telling him he should do more work on your idea
so you could get something you wanted without doing all of the work
for it.

And he did not respond favorably.

Wow, that's, uhm. Yeah, I guess that kind of thing happens sometimes.
I pretty much expect, in fact, that any time I tell someone to do unpaid
volunteer work on a task I value which they don't, they're going to
respond unfavorably.

-s

 

[Keith Thompson]

[nonsense snipped]

Whoops, nothing left!

 

[Keith Thompson]

[nonsense snipped]

Whoops, nothing left!

That was gratuitously snarky of me, and I probably should have kept
it to myself. Sorry about the noise.

 

[Dik T. Winter]

> Rui Maciel a écrit : ....
> Sure He treats me of "jerk" and I am not courteous enough.

You have the order reversed. First you were not courteous and in answer
to that he retorted in kind and called you a jerk.

 

[Phil Carmody]

[nonsense snipped]

Whoops, nothing left!

That was gratuitously snarky of me, and I probably should have kept
it to myself. Sorry about the noise.

We've certainly seen similar from Richard and others too. It's just
a slightly verbose, and fully justified, sigh. It proves you're a
human not a robot, don't worry about it. I'm happier knowing that I'm
conversing with humans.

(And mysteriously, the trolls seem to be closer to Markov models...)

Phil

 

[Nick Keighley]

Very courteous?  A message where you accuse Keith of "aren't always
just destrying" and "getting positive for a change"?  Saying things
like that to somebody is courteous?  Is that perhaps French courtesy?

Parisian?

 

[Nick Keighley]

I have to ask, yet again, why people keep feeding this troll. Is there
really value added to the newsgroup with these constant flamewars with
Mr. Navia?

I don't think he's a troll in that he intends to piss people off.
He *is* a compiler implementor and probably technically quite clued
in.
He sometimes has interesting technical points to make.
But he does like the sound of his own voice and he does hate
criticism,
constructive or not.

I'll be trying not to respond to him for a while.

 

[Default User]

I don't think he's a troll in that he intends to piss people off.

If you're saying that he doesn't intend to rile people up, then I think
I would disagree.

He is a compiler implementor and probably technically quite clued
in.

That doesn't prevent him from being a troll.

He sometimes has interesting technical points to make.

And it's fine to deal with him on that basis. These huge argumentive
threads provide little in the way of technical knowledge to the group.

I'll be trying not to respond to him for a while.

Brian

 

[Paul N]

I have never seen any serious proposal from thompson, heathfield
or any others of that clique to improve anything in the language.

As no-one seems to have said it, I will - you do appear to be right
here.

You are keen to see the C language develop, and have proposed a number
of changes to the language, which you have discussed here vigorously.

As far as I can tell, no-one else here is in favour of major changes
to the language. (I may of course be wrong about this, particularly as
I don't read comp.std.c which I believe is the correct place for
formal discussions about such changes.) Most people here seem in
favour of either leaving the language exactly as it is, or of only a
bit of mild tinkering. This seems to apply to everyone here - not just
"the regulars" but all the others who post here, regularly or not. I
don't recall seeing much discussion about why people are adverse to
change (though Richard H has said a little in response to your post
above). I presume it is that most people regard C as a "tight"
language, in which the source code can quickly and easily be converted
into compact, fast-running object code, and they worry that adding
extra "bells and whistles" to the language will detract from this. If,
on the other hand, you want to use some of the computer's power to
make life easier for the programmer (and why not? computers are a lot
bigger and faster than they used to be, there should be resources to
spare) and don't mind a bit of "bloat", then there are already
modifications to C that provide useful facilities, notably the
language C++, and people may feel that starting from there is better
than going back to C and striking out a new route from there.

So I don't think you can expect much in the way of support from other
people in the newsgroup for any major changes to the language that you
propose, and neither can you expect the others to be producing
exciting changes in the language of their own. Probably the best you
can expect is for people to comment on your suggestions; and they are
doing so.

I hope this calms things down a little...
Paul.

 

[jacob navia]

Paul N a écrit :

As no-one seems to have said it, I will - you do appear to be right
here.

You are keen to see the C language develop, and have proposed a number
of changes to the language, which you have discussed here vigorously.

As far as I can tell, no-one else here is in favour of major changes
to the language. (I may of course be wrong about this, particularly as
I don't read comp.std.c which I believe is the correct place for
formal discussions about such changes.) Most people here seem in
favour of either leaving the language exactly as it is, or of only a
bit of mild tinkering.

Not even the mild tinkering. I have often brought the example of the
asctime() function specification bug where an obvious overflow could
be avoided by dimensioning correctly a buffer or changing the specs
slightly.

I proposed (in comp.std.c) a change to the standard where a few lines
of code were modified.

Most people rejected the idea. asctime() overflows?

That's OK.

This seems to apply to everyone here - not just
"the regulars" but all the others who post here, regularly or not. I
don't recall seeing much discussion about why people are adverse to
change (though Richard H has said a little in response to your post
above). I presume it is that most people regard C as a "tight"
language, in which the source code can quickly and easily be converted
into compact, fast-running object code, and they worry that adding
extra "bells and whistles" to the language will detract from this.

Obviously you have a point here. But is this "performance" obsession
justified?

The overflow check that I proposed, for instance brings no measurable
change to the generated code and it will default to "off" so it is
not necessary to pay *anything* in performance terms to have this
optional feature.

Still, absolutely nothing.

If, on the other hand, you want to use some of the computer's power to
make life easier for the programmer (and why not? computers are a lot
bigger and faster than they used to be, there should be resources to
spare) and don't mind a bit of "bloat", then there are already
modifications to C that provide useful facilities, notably the
language C++, and people may feel that starting from there is better
than going back to C and striking out a new route from there.

The problem is that C++ took a route that has led to a language of
such a complexity that not even the creator of the language is able to
modify anything to it. See the "concepts" catastrophe when the
standards committee proposed to add some checking to template
arguments. Not even the creator of the language was able to develop
a solution within a few years time frame.

C, on the other hand, has retained its simplicity, what makes it a
better language because it is easier to learn and understand. My
proposed changes do not change anything in the language itself,
besides some syntactic sugar, or necessary "add-ons" like this overflow
proposal. The objective would be to make a language that is easier
to use (fewer "gotchas") but still simple and extremely fast.

The implementation of C I propose (and where I have implemented all the
changes) proves that it is possible to do it, and that the resulting
language is easier to use because the main problems like string handling
have been solved.

So I don't think you can expect much in the way of support from other
people in the newsgroup for any major changes to the language that you
propose, and neither can you expect the others to be producing
exciting changes in the language of their own.

Probably. What surprises me is that people (like thompson) that say that
a change would be *correct* and *necessary* (his words) answer in such
an emotional manner when asked to do the minimal effort of a proposal
to the committee.

The result is a committee that is in the same state of animated
suspension like many people in this group, and refuses to change
anything anywhere.

The end result is that C disappears from view. Most people now
think that C is an obsolete, dangerous language full of bugs.
And no, they do not even bother to write about it or to come here
to tell us that, they just do not care.

To put it with the words of Mr Glassborough:

"ACCU continues to have an interest in C but it is hard to demonstrate
that because people seem reluctant to write articles and reluctant to
propose C related presentations for its annual conference. "

Yes, nobody wants to be associated with something as obsolete as C.

This is not at all my opinion, and the reason I want to change things
is to preserve C as a valid language, and a valid option for programming
today.

Only by changing the language and proving it can evolve out of its
current mess (buffer overflows, bad string handling, obsolete library
etc) is there any chance of preserving it.

Probably the best you
can expect is for people to comment on your suggestions; and they are
doing so.

Well, maybe you are right and nothing can be done, C is already near
the grave anyway. I remember that troll that said (in this group) that
he was waiting for C programmers to die out. The average age here
is quite high, so no doubt if we go on like this just for a few years
there will be nobody left.

I hope this calms things down a little...
Paul.

Beyond this polemic however, is the sad fact that C is being destroyed
because is not allowed to evolve, both because the C++ people believe
they have created or are using a better language, and because the few
C programmers that are left cling to an absolute conservatism that makes
the problems of C even worse.

 

[Keith Thompson]

Paul N a écrit :

comp.std.c isn't the place for "formal discussions"; it has no
connection to the committee. But it's more appropriate than
comp.lang.c.

Most people here don't express an opinion on whether or how the
language should be changed. Most people here are trying to use the
language, and have neither the time nor the inclination to worry
about changes that, in the best case, won't show up in compilers
for a decade or more.

Not even the mild tinkering. I have often brought the example of the
asctime() function specification bug where an obvious overflow could
be avoided by dimensioning correctly a buffer or changing the specs
slightly.

I proposed (in comp.std.c) a change to the standard where a few lines
of code were modified.

Most people rejected the idea. asctime() overflows?

That's OK.

asctime() does have problems, but you have repeatedly overstated them.

gets() cannot be used safely (unless you have absolute control over
what appears on stdin), and I'm glad that it's being deprecated
and will eventually disappear from the language.

asctime() certainly can be used safely, because its misbehavior
occurs in response to arguments passed to it, which are under the
control of the programmer, not in response to uncontrolled input.
The argument to asctime() is a pointer to a struct tm; if that
struct tm was built from a valid time_t value, as it usually is,
then it's safe until the year 10000. It's only when you call
asctime() with a pointer to a struct tm that you've built yourself
that it can invoke undefined behavior -- and if you're doing that,
you'd better know what you're doing.

I dislike asctime(), and I never use it. I don't particularly like
what it does even when it's used correctly (its output format is a
US-centric one that I personally avoid in preference to YYYY-MM-DD,
and the trailing new-line is absurd). It exists in the language
only for historical reasons. I would support deprecating it and
eventually removing it from the language. I would also support
modifying its definition, even in ways that might change the behavior
of some conforming but contrived code, as long as that change of
behavior is explicitly acknowledged. I just don't think it's that
big a deal. I've heard of a grand total of 1 case where it caused
a problem; I think you, jacob, once ran into problems because you
used it incorrectly. (No offense intended; we all make mistakes.)

And I've said all this before, but apparently that doesn't count,
and you somehow conclude that everyone but you likes overflows.

[...]

The overflow check that I proposed, for instance brings no measurable
change to the generated code and it will default to "off" so it is
not necessary to pay *anything* in performance terms to have this
optional feature.

Still, absolutely nothing.

"Still, absolutely nothing"? What???

I have spent (and apparently wasted) substantial time discussing
your proposed changes in comp.lang.c and comp.std.c, pointing out
problems and suggesting improvements. If you don't like what I had
to say, that's fine. But don't pretend that there was no response.

comp.std.c has no official connection to the C standard committee;
most members probably don't even read Usenet. I understand that
submitting a proposal to the committee isn't easy -- but the fact
that these newsgroups are not the way to do it is not the fault of
anyone who participates here, and whining about it will not help
your cause.

[...]

Probably. What surprises me is that people (like thompson) that say that
a change would be *correct* and *necessary* (his words) answer in such
an emotional manner when asked to do the minimal effort of a proposal
to the committee.

I've explained why I'm not currently interested in making a proposal
to the committee as you suggest. My "emotional" response (calling
you an arrogant jerk) was in response to your personal insults,
not to what you asked me to do. And I don't believe I've used
the word "correct" or "necessary" in reference to your proposals;
please don't put words in my mouth.

Oh, and since you've told us before how much grief you've had to
to through when trying to make a proposal to the committee, I'm
surprised by your use of the phrase "minimal effort".

[snip]

 

[Keith Thompson]

"Still, absolutely nothing"? What???

I have spent (and apparently wasted) substantial time discussing
your proposed changes in comp.lang.c and comp.std.c, pointing out
problems and suggesting improvements. If you don't like what I had
to say, that's fine. But don't pretend that there was no response.

[...]

I should have mentioned that there were plenty of responses to your
proposal from other people, not just me.

 

[Richard Bos]

Parisian?

I dunno, the one Parisian I had a face-to-face conversation with was
remarkably civil. Then again, IIRC he wasn't _from_ Paris, he lived
there because of his job.

Richard

 

[Peter Nilsson]

Keith Thompson a écrit :

I think it is the first time thompson insults somebody.

Which should lead you to examine the cause more closely.

precisely when in a very courteous message,

You began an open letter with 'Look Keith:' In English,
that's quite an offensive way to start a letter. It can
be read as one step short of 'Look you arsehole.'

So, whilst it may or may not have been your intent to be
courteous, you have again begun a post by insulting the
very person (or people) you intend to persuade.

Keith was good enough to look past that and continue
reading. Of course, what followed later obviously cut
deeper.

Jacob, you are a querulant. Unfortunately, there's no easy
way of dealing with people exhibiting querulant behaviour.
Vexatious claims set up an environment of mutual hostility.

People who choose not to support you directly are not your
enemy, even (and especially) if they actually sympathise
and agree with you. They are simply people with other
priorities. They are entitled to their own priorities.
The sooner you realise and accept that the better.

 

[jacob navia]

Peter Nilsson a écrit :

You began an open letter with 'Look Keith:' In English,
that's quite an offensive way to start a letter. It can
be read as one step short of 'Look you arsehole.'

Look Mr Nilsson, surely I do not master all the english
language but just starting with "look keith" is not an
insult in any way, and you can't construct an insult where
there is none.

The problem is that I asked that guy to be positive and try
to make a proposal to modify the language

Accepting that would mean accepting that you can
improve the language and that C is not a fossilized
language of the past like the fossilized group here.

He dispelled immediately any misunderstanding by
*proving* that he belongs to the fossilized group of
programmers that dominates this discussion group.

He *needed* to insult to prove his membership to
that group.

Be it. He is now where he belongs.

I propose that we stop this anyway. My proposal was refused
and that was it. I really do not care a lot about what
that guy thinks of me. It was surprising that's all.

 

[Ben Bacarisse]

Peter Nilsson a écrit :

Look Mr Nilsson, surely I do not master all the english
language but just starting with "look keith" is not an
insult in any way, and you can't construct an insult where
there is none.

Your English is very good, but I don't see how you can argue this
point. That opening (which I see you have deliberately used again) is
rude. There are no two ways about it. I can certainly image that you
did not know that at the time, and it is clear from his explanation
the Keith passed over that, probably because he aware of how hard it
is to control tone in a language that is not your first.

The problem is that I asked that guy to be positive and try
to make a proposal to modify the language

Just a heads up: "that guy" is also impolite.

Accepting that would mean accepting that you can
improve the language and that C is not a fossilized
language of the past like the fossilized group here.

Accepting that would mean a lot of work. Not only in what Keith
suggested was missing (overflow on conversion) but your proposal about
overflow itself lacked the kind of technical rigour that a standard
needs. I can imagine that a linked proposal would require a lot of
work on both parts.

He dispelled immediately any misunderstanding by
*proving* that he belongs to the fossilized group of
programmers that dominates this discussion group.

You must have missed the messages where Keith has explained how he'd
like to see C develop. This is understandable if you don't read
everything posted, but if that is the case you should not make
assumptions about what people think.

<snip>

 

[Keith Thompson]

The problem is that I asked that guy to be positive and try
to make a proposal to modify the language

No, the problem is that you insulted me, and I had had enough.
After spending considerable time and effort trying to *help you*
to improve your own proposal, you challenged me to prove that I am
not "always just destroying".

Accepting that would mean accepting that you can
improve the language and that C is not a fossilized
language of the past like the fossilized group here.

No, accepting that would mean collaborating with you on a proposal
parallel to yours, something I'm not inclined to do. Do you
believe for a moment, given our respective attitudes, that such a
collaboration could work? And what exactly makes you think that
I'm under any obligation to do as you suggest?

He dispelled immediately any misunderstanding by
*proving* that he belongs to the fossilized group of
programmers that dominates this discussion group.

He *needed* to insult to prove his membership to
that group.

No, I insulted you in response to your insults directed at me.

jacob, if you read this far, stop and think for a moment.
Let's assume for the moment that you did not intend to insult me,
that your article that started this thread was intended to be
entirely positive. You've seen the reaction. You've been told
repeatedly that *I believed* that I had been personally insulted,
and multiple other people believed that supported me in that belief.

Did you spend even a moment considering that you might have actually
insulted me, even if it wasn't intentional? Did it even occur to
you that, writing in a language that isn't your first, you might
have gone too far?

If so many people tell me I'm wrong, my first thought is that
I just might be wrong, not that there's some massive conspiracy
against me.

Just think about it. I'm not even asking you to acknowledge anything
in public. Just go back to your original article, re-read it,
and *think about it*.

Be it. He is now where he belongs.

I propose that we stop this anyway. My proposal was refused
and that was it. I really do not care a lot about what
that guy thinks of me. It was surprising that's all.

You keep proposing that we stop this, but you seem unable to do so.
One way for this to stop is for you to stop posting on this thread.
I'm not telling you to stop, but if you want to, you can.

 

[wolfgang.riedel]

I'm not sure, but we are talking about signed Integer operations -
I think there might be OS's, that report loss of the significance of
sign bits, -
and those may not only happen by arithmetic operations, but also
by
unsigned shifts, logical or bit operations -
in a bit of a special register.
If so, you are free to provide an accessfunction,
otherwise - you are the compiler writer - I don't think, the language
must provide this -,
but you may in an extension or a library (Where to search after
something complex, look up PL/1).

Wolfgang