bug-gnulib
[Top][All Lists]
Advanced

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

Re: results with TinyCC/x86


From: Jeffrey Walton
Subject: Re: results with TinyCC/x86
Date: Wed, 3 Mar 2021 20:58:38 -0500

On Wed, Mar 3, 2021 at 8:37 PM Bruno Haible <bruno@clisp.org> wrote:
>
> On a x86_64 glibc system, building a Gnulib testdir with tcc (configured to
> produce x86 code: './configure --cpu=x86 --triplet=i386-linux-gnu') gives
> a number of test failures:
>
> FAIL: test-binary-io.sh
> FAIL: test-canonicalize
> FAIL: test-chown
> FAIL: test-closein.sh
> FAIL: test-fchownat
> FAIL: test-fclose
> FAIL: test-fdutimensat
> FAIL: test-fflush
> FAIL: test-fflush2.sh
> FAIL: test-fpurge
> FAIL: test-freading
> FAIL: test-freopen-safer
> FAIL: test-fseek.sh
> FAIL: test-fseeko.sh
> FAIL: test-fseeko3.sh
> FAIL: test-fseeko4.sh
> FAIL: test-fstatat
> FAIL: test-ftell3
> FAIL: test-ftello.sh
> FAIL: test-ftello2.sh
> FAIL: test-ftello3
> FAIL: test-ftello4.sh
> FAIL: test-futimens
> FAIL: test-get-rusage-as
> FAIL: test-lchown
> FAIL: test-linkat
> FAIL: test-lseek.sh
> FAIL: test-lstat
> FAIL: test-pread.sh
> FAIL: test-pwrite.sh
> FAIL: test-read-file
> FAIL: test-rename
> FAIL: test-renameat
> FAIL: test-renameatu
> FAIL: test-stat
> FAIL: test-statat
> FAIL: test-supersede
> FAIL: test-truncate
> FAIL: test-utime
> FAIL: test-utimens
> FAIL: test-utimensat
> FAIL: test-yesno.sh
> FAIL: test-fts
>
> The standard I/O functions appear to be severely broken in this configuration.
> (My guess is that tcc does not support the symbol versioning used by glibc.)

Before you go down a rabbit hole on this...

Does GNU have a policy in place for supporting compilers?

In the past I've seen it stated in terms of market share. For example,
the United States IRS supports browsers that have 5% or greater market
share. The IRS would support IE, Forefox, Opera and several others
based on statistics gathered from their web logs.

Some systems you won't be able to gather statistics. For example, you
won't know the market share of the SunCC compiler on Solaris or XLC on
AIX. For SunCC or XLC you should assume it meets policy requirements
because it is the vendor's native compiler.

I believe Debian collects statistics on program usage. Maybe you
should look at TinyCC usage before you decide to spend a lot of time
on it.

Jeff



reply via email to

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