tinycc-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Tinycc-devel] mob-stuff branch?


From: grischka
Subject: Re: [Tinycc-devel] mob-stuff branch?
Date: Tue, 25 Oct 2022 13:34:43 +0200
User-agent: Mozilla/5.0 (Windows NT 6.0; rv:38.0) Gecko/20100101 Thunderbird/38.5.0

On 25.10.2022 00:08, Bernhard Reutner-Fischer wrote:
AFAICS the "warn about incorrect use of output_*" by using an enum did
never end up on mob.

Just because features often come with more complexity does not mean
that more complexity already is a feature.

Much more often the contrary is true:  the simpler the better.

"int output_type;" is simple.

"output_t output_type;" is "output_t(ype)" twice.  Honestly, I still
fail to see the point, even 13 years later ... ;)

-- gr


We seem to have --{no-,}whole-archive by now, so that's obsolete i
guess. And the elf-interpreter toggle is in there, too, so that's fine.

Not sure what you would think about the idea i toyed with in "add
--enable-shared, part1". IIRC the idea was to cut down on memory usage
when running many tcc processes or something along those lines.

thanks!

$ git log -n7 origin/mob-stuff
commit bfa394dab88cb417bb540c32647f04d08e9838af (origin/mob-stuff)
Author: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Date:   Wed Sep 2 15:55:48 2009 +0200

      Move bounds-checking code to a bcheck.a

      ... to avoid undefined references to __bound_new_region()
      (when using a libtcc1_s.so instead of libtcc1.a) on i386.


That's an open question I think.


      Signed-off-by: aldot <rep.dot.nop@gmail.com>

commit a39d055d0e666b152df87b4a603c670c3af37d68
Author: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Date:   Wed Sep 2 15:55:48 2009 +0200

      Handle --whole-archive

      Support and document -Wl,--{no-,}whole-archive.

done.
Didn't check if what's in there actually works.


      Signed-off-by: aldot <rep.dot.nop@gmail.com>

commit f0e8b23d02a4c06789b4b8876c65866a237ce570
Author: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Date:   Tue Sep 1 13:32:34 2009 +0200

      Implement -m{uclibc,glibc} to choose elf_interp

        Support -muclibc and -mglibc (default) to choose between the elf
        interpreter like gcc does.

        $ ./tcc -o hi-glibc hello.c
        $ ./tcc -muclibc -o hi-uclibc hello.c
        $ readelf -l hi-* | egrep "(^File: | interpreter)"
      File: hi-glibc
            [Requesting program interpreter: /lib/ld-linux.so.2]
      File: hi-uclibc
            [Requesting program interpreter: /lib/ld-uClibc.so.0]

Presumably done.
Didn't check how it's implemented though.


      Signed-off-by: aldot <rep.dot.nop@gmail.com>

commit 83cb1364a606fa9fbff450a65ab68e09a5bb4e42
Author: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Date:   Wed Sep 2 12:03:51 2009 +0200

      Revert "add --enable-shared, part1"

      This reverts commit 9257c5221ea4f095d813082716f786c6c5d23321.

      Signed-off-by: aldot <rep.dot.nop@gmail.com>

commit d90eb2008b62d19f70bdd14bdf9d784b8789da5d
Author: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Date:   Tue Sep 1 17:04:42 2009 +0200

      add --enable-shared, part1

        Build a libtcc.so and link the tcc binary against it.

That area is, i believe, not dealt with yet, is it?

It would serve 2 purposes:
- Runtime mem consumption reduction.
- Would help establish an improved interface to the "backend".

I'd encourage discussion around that area.
Mainly to cleanup an streamline internal interfaces TBH.


      Signed-off-by: aldot <rep.dot.nop@gmail.com>

commit 8a555345e3faa37ff0b2c0b84ae1f42358400923
Author: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Date:   Tue Sep 1 14:27:41 2009 +0200

      warn about incorrect use of output_*


Not addressed yet.
I think we should do something along those lines. WDYT?


      Signed-off-by: aldot <rep.dot.nop@gmail.com>

commit 3a40b44938d59b71d0c010579d9bd9f05e7eed46
Author: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Date:   Tue Sep 1 11:52:51 2009 +0200

      allow for non-stripped install


That's not dealt with as far as I'm aware.
Nowadays distros tend to split off -dbgsym packages on their own. So if we 
strip the debug on our own, this defeats usability and debug ability for distro 
users IMO.

I think we should at least give distros an easy way to decide on their own, 
shouldn't we.

thanks and cheers,





reply via email to

[Prev in Thread] Current Thread [Next in Thread]