Re: testsuite on Linux

[Benjamin Kosnik <bkoz@cygnus.com>]

> Here are the results from 'make check-install' with current CVS soruces.
> Only one file failed to build and two failed to pass; I've pasted the
> contents of the logfile at that point.  Things look pretty good.

Cool this is what I get as well. It's so nice to see somebody post 
results that match mine, I was getting a bit worried.

> (The ostream_streambuf test never made it into CVS.  Is it pretty safe to
> assume that the relevent function is going to work, so that I can take those
> three files out of my working directory?)

I still am hanging onto your test case. I'm half-way through the 
stringbuf changes, and hope to have the rest done today/tomorrow, 
depending. I'm hoping to make a more elaborate test case, based on yours, 
okay? It's not forgotten.

> 44	fail	367662	33072	2672844	21_strings/inserters_extractors.cc

This includes a test for ostringstreams, thus the failure. Everything 
else (istreams extractor, getline) seem to work, so this will pass as 
soon as I check in the stringstream changes.

> 1	fail	0	0	0	26_numerics/modf_float.cc
> ../libstdc++/testsuite/26_numerics/modf_float.cc: In function `void test()':
> ../libstdc++/testsuite/26_numerics/modf_float.cc:30: passing `float *' as
> argument 2 of `modf(double, double *)'

This is broken for me too--it's something to do with the autoconf/math 
intrinsics that Gaby was working on. I don't know what the status is on 
this, but will see what Gaby says.

> 68	fail	376478	32656	2756682	27_io/stringbuf.cc
> stringbuf.st-exe: ../libstdc++/testsuite/27_io/stringbuf.cc:78: bool
> test03():

I broke this on friday, and will be fixed as I check in the remaining 
stringbuf fixes.