Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.c > #176635 > unrolled thread

MCC Compiler

Back to article view | Back to comp.lang.c

Contents

  MCC Compiler Bart <bc@freeuk.com> - 2023-09-28 12:14 +0100
    Re: MCC Compiler Bart <bc@freeuk.com> - 2023-09-28 15:31 +0100
    Re: MCC Compiler Tim Rentsch <tr.17687@z991.linuxsc.com> - 2023-09-28 15:50 -0700
      Re: MCC Compiler Bart <bc@freeuk.com> - 2023-09-29 00:24 +0100
        Re: MCC Compiler Kaz Kylheku <864-117-4973@kylheku.com> - 2023-09-29 00:38 +0000
        Re: MCC Compiler Tim Rentsch <tr.17687@z991.linuxsc.com> - 2023-09-28 22:10 -0700
          Re: MCC Compiler Bart <bc@freeuk.com> - 2023-09-29 11:21 +0100
            Re: MCC Compiler Malcolm McLean <malcolm.arthur.mclean@gmail.com> - 2023-09-29 03:35 -0700
              Re: MCC Compiler Bart <bc@freeuk.com> - 2023-09-29 11:59 +0100
                Re: MCC Compiler Michael S <already5chosen@yahoo.com> - 2023-09-29 04:07 -0700
                  Re: MCC Compiler Malcolm McLean <malcolm.arthur.mclean@gmail.com> - 2023-09-29 04:36 -0700
                Re: MCC Compiler gazelle@shell.xmission.com (Kenny McCormack) - 2023-09-29 11:57 +0000
                Re: MCC Compiler David Brown <david.brown@hesbynett.no> - 2023-09-29 16:40 +0200
                  Re: MCC Compiler Bart <bc@freeuk.com> - 2023-09-29 16:46 +0100
              Re: MCC Compiler gazelle@shell.xmission.com (Kenny McCormack) - 2023-09-29 11:55 +0000
                Re: MCC Compiler Anton Shepelev <anton.txt@gmail.moc> - 2023-09-29 18:18 +0300
                  Re: MCC Compiler Tim Rentsch <tr.17687@z991.linuxsc.com> - 2023-10-03 08:21 -0700
              Re: MCC Compiler Richard Harnden <richard.nospam@gmail.com> - 2023-09-29 15:47 +0100
            Re: MCC Compiler Michael S <already5chosen@yahoo.com> - 2023-09-29 04:06 -0700
          Re: MCC Compiler Bart <bc@freeuk.com> - 2023-09-29 11:48 +0100

#176635 — MCC Compiler

My MCC project was intended to upgrade my BCC C compiler to use a 
similar backend to my other compilers.

The aims were:

* A more solid code generator that would run more C programs with
   fewer errors

* To apply the modest optimiser of that new backend

* To eliminate the intermediate ASM stages of BCC and directly
   generate EXEs

* To have faster compilation because of that

* To have full ABI compliance, with proper handling of struct
   passing

As it turned out, only the last of those has been achieved, since:

* The generated code, even if better, did not magically make C
   programs work that didn't before. It is impractical to debug
   large convoluted C programs to find a bug that exists elsewhere
   entirely

* I started the direct EXE backend yesterday, but tedium had set in
   and I abandoned it. So MCC is slower than BCC, partly because of the
   extra IL stage, but also the ASM stages are handled discretely. In
   BCC, the ASMs were kept in memory and the assembler was incorporated.

* BCC was a one-file solution (compiler + headers + assembler/linker).
   MCC was intended as a 2-file one (windows.h was taken outside; it's
   too big). As it is, MCC is a 3-file solution (mcc.exe + aa.exe +
   windows.h) which also makes it slower as I said.

* I haven't bothered activating the optimiser.

* I found out that, even without the final EXE generation, it was still
   slower than Tiny C, when there are lots of smaller files (with bigger
   files, the speed would be about par I think)

So, MCC is still a usable product, and would be a better starting point 
for further work (for example, to create an interpreter for its IL, 
which opens up debugging capabilities), but I now have two private, 
mediocre C compilers.

Something I added to both was support for wildcard inputs like '*.c'. I 
did this to more easily build projects like Malcolm's BBX which could be 
done like this:

     mcc *.c freetype\* samplerate\*.c

To avoid having to type that lot, I put those inputs into an '@' options 
file called 'bbx':

     *.c freetype\* samplerate\*.c

Now I just type 'mcc @bbx'. I was surprised however when I tried to do:

     gcc @bbx

because:

(1) gcc doesn't like backslashes in options files; they have to be "/"

(2) gcc doesn't expand wildcard filespecs like "*.c" inside options
     files

Why not? Since this is not done by the OS, it must be done by the app 
anyway.

But that's one more small thing at least that my compilers do better!

(The bigger projects I've used as test inputs were:

   Lua
   TCC
   Pico C
   SQL (sqlite3.c + shell.c)
   BBX
   LibJPEG
   AA, QQ, CC (transpiled C from three of my projects)

The last three seem to run fine (my conservative C code always does!). 
Most of the others seemed to work based on superficial tests, but I 
didn't do in-depth testing.

But TCC didn't work. Version 0.9.27 (I tried also .25 and .28) went the 
furthest: it could compile programs and produce EXEs, but the EXEs were 
faulty.

At some point I will have another go at Seed7.)

[toc] | [next] | [standalone]

#176653

On 28/09/2023 12:14, Bart wrote:
> My MCC project was intended to upgrade my BCC C compiler to use a 
> similar backend to my other compilers.

> But TCC didn't work. Version 0.9.27 (I tried also .25 and .28) went the 
> furthest: it could compile programs and produce EXEs, but the EXEs were 
> faulty.

MCC can generate ASM source code with named locals and temps, or using 
only their offsets to keep the size down, which is what I'd been using 
but had forgotten about.

But there was a problem with the offsets for temps (used when I run out 
of registers) which I have to look at.

Generating ASMs with fully named temps, then Tiny C works. Further, the 
exe produced can compile itself, to multiple generations. (BCC only 
managed one generation.)

So it's not quite as poor as I'd thought. I might have a go at 0.9.28 again.

[toc] | [prev] | [next] | [standalone]

#176678

Bart <bc@freeuk.com> writes:

> My MCC project was intended to upgrade my BCC C compiler to use a
> similar backend to my other compilers.  [...]

Please confine your postings in comp.lang.c to topics and subjects
relevant to the C language.  None of what you say in your posting
is topical in comp.lang.c.  An obvious suggestion is the newsgroup
comp.compilers instead.

[toc] | [prev] | [next] | [standalone]

#176684

On 28/09/2023 23:50, Tim Rentsch wrote:
> Bart <bc@freeuk.com> writes:
> 
>> My MCC project was intended to upgrade my BCC C compiler to use a
>> similar backend to my other compilers.  [...]
> 
> Please confine your postings in comp.lang.c to topics and subjects
> relevant to the C language.  None of what you say in your posting
> is topical in comp.lang.c.  An obvious suggestion is the newsgroup
> comp.compilers instead.

That's me told!

This project was first mentioned as a possibility at the end of August. 
I was giving an update on it before moving on to my normal projects and 
saner languages.

Maybe some people here (obviously not the old-timers) would find a blog 
report on the tribulations of creating a personal C compiler interesting.

Or maybe someone is thinking of doing something of their own, and my 
experience shows one-man compilers can be viable.

So how about letting them make up their own mind. If not, it is easy 
enough to ignore the post or to killfile me.

Meanwhile I could equally suggest you post all your deadly-dull stuff 
about C standards minutiae to comp.std.c

[toc] | [prev] | [next] | [standalone]

#176688

On 2023-09-28, Bart <bc@freeuk.com> wrote:
> On 28/09/2023 23:50, Tim Rentsch wrote:
>> Bart <bc@freeuk.com> writes:
>> 
>>> My MCC project was intended to upgrade my BCC C compiler to use a
>>> similar backend to my other compilers.  [...]
>> 
>> Please confine your postings in comp.lang.c to topics and subjects
>> relevant to the C language.  None of what you say in your posting
>> is topical in comp.lang.c.  An obvious suggestion is the newsgroup
>> comp.compilers instead.
>
> That's me told!

comp.compilers *is* a nice newsgroup, though.

Mind you, it's moderated.

You can't go on long C sucks tirades and whatnot.

[toc] | [prev] | [next] | [standalone]

#176704

Bart <bc@freeuk.com> writes:

> On 28/09/2023 23:50, Tim Rentsch wrote:
>
>> Bart <bc@freeuk.com> writes:
>>
>>> My MCC project was intended to upgrade my BCC C compiler to use a
>>> similar backend to my other compilers.  [...]
>>
>> Please confine your postings in comp.lang.c to topics and subjects
>> relevant to the C language.  None of what you say in your posting
>> is topical in comp.lang.c.  An obvious suggestion is the newsgroup
>> comp.compilers instead.
>
> That's me told!
>
> This project was first mentioned as a possibility at the end of
> August.  I was giving an update on it before moving on to my normal
> projects and saner languages.
>
> Maybe some people here (obviously not the old-timers) would find a
> blog report on the tribulations of creating a personal C compiler
> interesting.
>
> Or maybe someone is thinking of doing something of their own, and my
> experience shows one-man compilers can be viable.

None of that changes the reality that your comments are not
about the C language or C programs and so are not topical here.

> So how about letting them make up their own mind.  If not, it is easy
> enough to ignore the post or to killfile me.

I know that your MO is to be a rude, self-centered, insecure
jerk.  You have been for years and there is no indication that
will change any time soon.  There is plenty of evidence that
people in the newgroup here are not interested in what you have
to say about your experience working on compilers, especially
since those compilers are not written in C and aren't faithful to
the C language (and it isn't clear whether you don't know that or
if you just don't care).  So people have made up there minds;
the only question is how long it will take you to realize it, if
ever.

> Meanwhile I could equally suggest you post all your deadly-dull stuff
> about C standards minutiae to comp.std.c

My comments about the details of C pertain to the C language and
how to write C programs.  They are topical here.  Discussions in
comp.std.c are supposed to do with how to change the C standard
or how to read the C standard in areas where there is some
question about what is meant.  If you think my comments are
deadly dull you are free to ignore the postings or put me in
your kill file.  But whether you do that or not, my comments
are topical, and yours are not.

[toc] | [prev] | [next] | [standalone]

#176713

On 29/09/2023 06:10, Tim Rentsch wrote:

> I know that your MO is to be a rude, self-centered, insecure
> jerk.

What's yours?

[toc] | [prev] | [next] | [standalone]

#176714

On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote:
> On 29/09/2023 06:10, Tim Rentsch wrote: 
> 
> > I know that your MO is to be a rude, self-centered, insecure 
> > jerk.
> What's yours?
>
Oh don't lower the tone by replying to stuff like that.

I think Tim is in a mood. He posted something negative about me as
well. It's not normal for him.
 

[toc] | [prev] | [next] | [standalone]

#176716

On 29/09/2023 11:35, Malcolm McLean wrote:
> On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote:
>> On 29/09/2023 06:10, Tim Rentsch wrote:
>>
>>> I know that your MO is to be a rude, self-centered, insecure
>>> jerk.
>> What's yours?
>>
> Oh don't lower the tone by replying to stuff like that.
> 
> I think Tim is in a mood. He posted something negative about me as
> well. It's not normal for him.
>   

I think it is!

He seems to like policing this /unmoderated/ newsgroup according to what 
/he/ finds interesting or relevant.

Meanwhile many here are happy to participate in endless discussions 
about the 'halting' problem.

[toc] | [prev] | [next] | [standalone]

#176718

On Friday, September 29, 2023 at 1:59:18 PM UTC+3, Bart wrote:
> On 29/09/2023 11:35, Malcolm McLean wrote: 
> > On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote: 
> >> On 29/09/2023 06:10, Tim Rentsch wrote: 
> >> 
> >>> I know that your MO is to be a rude, self-centered, insecure 
> >>> jerk. 
> >> What's yours? 
> >> 
> > Oh don't lower the tone by replying to stuff like that. 
> > 
> > I think Tim is in a mood. He posted something negative about me as 
> > well. It's not normal for him. 
> >
> I think it is! 
> 
> He seems to like policing this /unmoderated/ newsgroup according to what 
> /he/ finds interesting or relevant. 
> 
> Meanwhile many here are happy to participate in endless discussions 
> about the 'halting' problem.

Many? You are wildly exaggerating.

[toc] | [prev] | [next] | [standalone]

#176720

On Friday, 29 September 2023 at 12:07:50 UTC+1, Michael S wrote:
> On Friday, September 29, 2023 at 1:59:18 PM UTC+3, Bart wrote: 
> > On 29/09/2023 11:35, Malcolm McLean wrote: 
> > > On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote: 
> > >> On 29/09/2023 06:10, Tim Rentsch wrote: 
> > >> 
> > >>> I know that your MO is to be a rude, self-centered, insecure 
> > >>> jerk. 
> > >> What's yours? 
> > >> 
> > > Oh don't lower the tone by replying to stuff like that. 
> > > 
> > > I think Tim is in a mood. He posted something negative about me as 
> > > well. It's not normal for him. 
> > > 
> > I think it is! 
> > 
> > He seems to like policing this /unmoderated/ newsgroup according to what 
> > /he/ finds interesting or relevant. 
> > 
> > Meanwhile many here are happy to participate in endless discussions 
> > about the 'halting' problem.
> Many? You are wildly exaggerating.
>
There's one poster who occasionally posts to comp.lang.c who believes he has
found a flaw in the straightforwards proof that halting is non-computable that
everyone knows (run the halt detector on itself then loop forever if it detects
halt, and halt if it detects non-halting behaviour is the gist).  The volume of 
discussion that this has created is unbelieveable and, yes, I have participated 
myself.

Mostly he posts to comp.theory however, where it is at least technically on topic.

[toc] | [prev] | [next] | [standalone]

#176722

In article <uf6alm$809b$2@dont-email.me>, Bart  <bc@freeuk.com> wrote:
>On 29/09/2023 11:35, Malcolm McLean wrote:
>> On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote:
>>> On 29/09/2023 06:10, Tim Rentsch wrote:
>>>
>>>> I know that your MO is to be a rude, self-centered, insecure
>>>> jerk.
>>> What's yours?
>>>
>> Oh don't lower the tone by replying to stuff like that.
>> 
>> I think Tim is in a mood. He posted something negative about me as
>> well. It's not normal for him.
>>   
>
>I think it is!

Yes.  It is.

>He seems to like policing this /unmoderated/ newsgroup according to what 
>/he/ finds interesting or relevant.

Yup.

>Meanwhile many here are happy to participate in endless discussions 
>about the 'halting' problem.

Yup.

The rules are obviously applied, to put it charitably, inconsistently.

Topicality rules for thee, but not for me.

-- 
Pensacola - the thinking man's drink.

[toc] | [prev] | [next] | [standalone]

#176733

On 29/09/2023 12:59, Bart wrote:
> On 29/09/2023 11:35, Malcolm McLean wrote:
>> On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote:
>>> On 29/09/2023 06:10, Tim Rentsch wrote:
>>>
>>>> I know that your MO is to be a rude, self-centered, insecure
>>>> jerk.
>>> What's yours?
>>>
>> Oh don't lower the tone by replying to stuff like that.
>>
>> I think Tim is in a mood. He posted something negative about me as
>> well. It's not normal for him.
> 
> I think it is!
> 
> He seems to like policing this /unmoderated/ newsgroup according to what 
> /he/ finds interesting or relevant.
> 

Tim is an extraordinarily patronising grumpy old git.  (And that is 
coming from someone who knows he can often be patronising!)  He is also 
very rarely wrong about the technicalities of C, and thus a useful 
resource for this group.

I don't think you need fear the consequences of being on his bad side - 
he killfiled me years ago, without it doing me any harm.

I do appreciate that it's useful for there to be occasional "policing" 
of the group - polite reminders when threads are straying too far from 
topicality.  We want to keep the group strongly focused on C.  (And I 
say that as someone who is on the receiving end of such reminders every 
now and again - I don't always follow them immediately, but I approve of 
the principle.)

Discussions about specific compiler details have traditionally been 
off-topic in comp.lang.* groups - its too easy to get bogged down by 
compiler-specific threads rather than threads about the language itself. 
  Some mention of common compilers is inevitable.

However, I do think that a thread about your C compiler is very much 
on-topic.  You might get into a heated discussion, but that's hardly 
unusual.  You might get tips or ideas, or answers to questions.


> Meanwhile many here are happy to participate in endless discussions 
> about the 'halting' problem.

Actually, almost all the comp.lang.c regulars were strongly against such 
discussions, at least when it became clear that the person behind them 
was off his head and not the slightest bit interested in listening to 
rational arguments or mathematical realities.

But "whataboutism" is not a good argument anyway.

[toc] | [prev] | [next] | [standalone]

#176744

On 29/09/2023 15:40, David Brown wrote:

> However, I do think that a thread about your C compiler is very much 
> on-topic.  You might get into a heated discussion, but that's hardly 
> unusual.  You might get tips or ideas, or answers to questions.

You might be pleased to know that the revised version is less verbose by 
default when more than one file is involved.

But it is still somewhat more so than gcc:

    c:\luac>gcc @lua

    c:\luac>mcc @lua
    Compiling 33 files to lua.exe

[toc] | [prev] | [next] | [standalone]

#176721

In article <2c2f6ab4-949a-46ff-b97c-036103f396d5n@googlegroups.com>,
Malcolm McLean  <malcolm.arthur.mclean@gmail.com> wrote:
>On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote:
>> On 29/09/2023 06:10, Tim Rentsch wrote: 
>> 
>> > I know that your MO is to be a rude, self-centered, insecure 
>> > jerk.
>> What's yours?
>>
>Oh don't lower the tone by replying to stuff like that.
>
>I think Tim is in a mood. He posted something negative about me as
>well. It's not normal for him.
> 

It *is* normal for him.

That's all he does - posts the most boring, uninteresting drivel
imaginable, but, yet, at least by the looser standard now generally used in
CLC, on-topic.

(Note and clarification of the above: By the usual standard of most groups,
as applied to the specific rules of CLC, *nothing* is on-topic here.  But
of course, that standard is rarely applied, since it would lead to, well,
you know, nothing ever being posted)

-- 
Just like Donald Trump today, Jesus Christ had a Messiah complex.

And, in fact, the similarities between the two figures are quite striking.
For example, both have a ragtag band of followers, whose faith cannot be shaken.

[toc] | [prev] | [next] | [standalone]

#176742

Kenny McCormack about Tim:

> That's all he does - posts the most boring, uninteresting
> drivel imaginable, but, yet, at least by the looser
> standard now generally used in CLC, on-topic.

Technical discussions can be boring to an observer who has
mastered the subject, but I for one appreciate Tim's
feedback to my articles.

-- 
()  ascii ribbon campaign -- against html e-mail
/\  www.asciiribbon.org   -- against proprietary attachments

[toc] | [prev] | [next] | [standalone]

#177251

Anton Shepelev <anton.txt@gmail.moc> writes:

> Kenny McCormack about Tim:
>
>> That's all he does - posts the most boring, uninteresting
>> drivel imaginable, but, yet, at least by the looser
>> standard now generally used in CLC, on-topic.

Apparently Kenny has forgotten a recent posting of mine,
responding to his posting asking something, that was meant
to provide a helpful answer to his question.

> Technical discussions can be boring to an observer who has
> mastered the subject, but I for one appreciate Tim's
> feedback to my articles.

Thank you, it's nice to hear that someone appreciates my
comments.

[toc] | [prev] | [next] | [standalone]

#176736

On 29/09/2023 11:35, Malcolm McLean wrote:
> On Friday, 29 September 2023 at 11:21:44 UTC+1, Bart wrote:
>> On 29/09/2023 06:10, Tim Rentsch wrote:
>>
>>> I know that your MO is to be a rude, self-centered, insecure
>>> jerk.
>> What's yours?
>>
> Oh don't lower the tone by replying to stuff like that.
> 
> I think Tim is in a mood. He posted something negative about me as
> well. It's not normal for him.
>   

Perhaps ...

He can tell intuitively that 7 things annoy him, but he can only be 
concurrently irritated about 3 of them.

...?

No?  I'll get my coat.

[toc] | [prev] | [next] | [standalone]

#176717

On Friday, September 29, 2023 at 1:21:44 PM UTC+3, Bart wrote:
> On 29/09/2023 06:10, Tim Rentsch wrote: 
> 
> > I know that your MO is to be a rude, self-centered, insecure 
> > jerk.
> What's yours?

Modus operandi. 
Is there kitchen.lang.latin newsgroup?

[toc] | [prev] | [next] | [standalone]

#176715

On 29/09/2023 06:10, Tim Rentsch wrote:
> Bart <bc@freeuk.com> writes:

> to say about your experience working on compilers, especially
> since those compilers are not written in C

Is that really relevant? Does a compiler for language L have to be 
written in L to be taken seriously?

In any case in can be transpiled to C in 60ms:

   c:\cx>tm mc -c cc
   Compiling cc.m---------- to cc.c
   TM: 0.06

> and aren't faithful to
> the C language (and it isn't clear whether you don't know that or
> if you just don't care).

In which ways? My product compiles a C 'subset' but does not formally 
define what it is. Yet it manages to compile 100s of 1000s of lines of C 
applications (not my generated code) which would be challenging for many 
such small compilers.

But it no longer supports my own extensions, like the many that gcc 
provides, and it no longer tries to improve on the language or fail 
programs on features I feel ought to be deprecated.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.c

csiph-web