On 8/16/2010 4:51 PM, Peter Rosin wrote: > I don't think it's a race, I think it's from the code below: > > func_verbose "creating GNU ld script: $output" > > (approx line 7070 in ltmain.m4sh) > > where an absolute file name is written to the linker script > and therefore escapes the MSYS argument mangling.
Ah. And that code is only hit when you exceed the max command line length -- which explains why the test passes "normally". I guess the reason I'm just now noticing this is because until now, libtool on MinGW/MSYS had several test failures in "normal" mode, and I just *assumed* that those failures were the reason the max_command_length test failed. But, obviously, this failure was ALSO hiding in there all along. > Will be > easily fixed by your unmerged patches if you haven't fixed > it already (which I would think you have)... I don't recall if I fixed that particular issue or not, but you're right -- with my unmerged patch it's trivial to fix this issue. So, we can wait for that eventuality (it's first up in my personal queue once this whole sysroot thing is merged -- or, at minimum, more-or-less finalized within the sysroot branch). -- Chuck