Re: [RFC] make check-abi

[Phil Edwards <phil at jaj dot com>]

On Tue, Aug 20, 2002 at 08:53:07PM -0700, Benjamin Kosnik wrote:
> 1) where do the baseline files go?? The hooks are in the current patch
>    to make use of the configure bits. From Loren's comments, it looks
>    like 3.2.0 baselines will be necessary for every unique target that
>    attempts to do abi checking. If that's the case, might as well do
>    object size checking by default anyway.

config?  config/abi?  config/abitest?

Then baselines can go in a file named after the platform, e.g.,
config/abi-3.2/<target_triplet>, and we can point a variable there using
the last stanza in configure.target if we need to override choices.


> 2) how does one get the current library.so name?
[...]
> + ## XXX Fix me.
> + test_library = ${top_builddir}/src/.libs/libstdc++.so.5.0.1
> + test_symbols: ${test_library}
> + 	@if [ ! -f ${test_file} ]; then \
> + 	   readelf -s -W ${test_library} |

We could either parse the libtool_VERSION varaible, or just use libstdc++.so.
For Linux it follows the symlink (that's what my scripts are using),
for BSD the file is named libstdc++.so anyway.


I am renumbering the...
> Other issues:
> 
> 3) should system() be used instead of automake-direct-quoting insanity?

That's my vote.  There are going to be problems eventually either way
(requiring GNU binutils), so whether we make the call from compiled code
or from scripts makes no difference.


> 4) the testsuite/* directory is starting to get a bunch more stuff in
> it. Should testsuite/util (or whatever) be started? I'm not quite sure
> this is a big issue, or one that has to be addressed right now, but
> still. Curious.

I would hold off a bit longer, but I don't feel strongly about it.


Phil

-- 
I would therefore like to posit that computing's central challenge, viz. "How
not to make a mess of it," has /not/ been met.
                                                 - Edsger Dijkstra, 1930-2002