This is the mail archive of the
mailing list for the GCC project.
Re: a question about fix-header.c
Once upon a time, long ago and far away, I moved the "fixincludes"
script from the gcc directory into a subdirectory named, "fixinc".
The current TOT has that subdirectory moved into a quasi-independent
project directory at the same level as the main "gcc" directory.
This directory is named, "fixincludes". "fix-header.c" fixes headers
just like "fixincludes", but in a much different way and it _still_
lives in the gcc directory. Its job is to ensure that POSIX defined
interfaces are in their correct header files. "free" is among the
interfaces it wants to ensure exists in stdlib.h. fix-header and
fixincludes don't cooperate very well. It would be good to munge
any useful stuff fix-header does into fixincludes.
Kenneth Zadeck wrote:
> Having just tortured danny berlin, the person who started me on this
> quest, I can now say "never mind". He had pointed me to some code that
> is most likely unused but still lives in the gcc source directory that
> purportes to fix include files.
> I have now been told that there is a version of your project that lives
> either lives in or at least the output lives in some other part of the
> gcc directories. I will start looking there and only bother you again
> if this turns out to also be wrong.
> Thankyou for your time.
> Sorry again.
> Bruce Korb wrote:
> >Bruce Korb wrote:
> >>Kenneth Zadeck wrote:
> >>>I would like to modify fix includes so that it adds the attribute
> >>>"pointer-no-escape" onto the function definition of free in stdlib.h
> >>>(and possibly some other fuctions as well).
> >>>I am having trouble figuring out how to do this.
> >>>Dan Berlin said that you might be able to point me in the
> >>correct direction.
> >"correct direction" ?? Maybe I should have remembered the title,
> >instead of reading, "modify fix includes" too literally.
> >"fix-header.c" is a different program that is included in the
> >more-or-less same process as the "fixincl" program that I wrote.
> >For aesthetic reasons, it would be nice to incorporate its
> >functionality into the "fixincl" program. I looked at doing that
> >a few years ago, but determined it was more effort than it
> >was worth (to me). At this point, now that fixincludes is
> >a semi-autonomous installable project, it would be a good thing
> >to do. Not today. Meanwhile, as far as "correct direction" is
> >concerned, I'd have to "read the source, Luke" to try to
> >re-understand what fix-header was doing. Sorry. :(
> >Regards, Bruce