https://sourceware.org/bugzilla/show_bug.cgi?id=25713
--- Comment #29 from Torbjörn SVENSSON <torbjorn.svensson at st dot com> --- (In reply to Fred Eisele from comment #28) > However, I do not think PR 25713 is going to tackle that larger issue. > I think it should be restricted to dealing with the filename as provided by > the first argument to _bfd_real_fopen. > https://github.com/bminor/binutils-gdb/blob/ > fb583cfc0936554a1f8d86ef2b6649bf511d6df8/bfd/bfdio.c#L89 > > FILE *_bfd_real_fopen (const char *filename, const char *modes); I agree that this PR can't handle all the cases, I'm just highlighting that the problem is wider than what you might think at a first glance. The "filename" parameter to _bfd_real_fopen(), isn't that in the native encoding provided by the user? If so, it do have the problem I'm pointing to in comment 27. Keep in mind that I'm not stating that we need to address the problem, I'm just saying that it would give unexpected results in the situation of a character outside the ASCII set. > Is there any plan to tackle the issue of accommodating paths from multiple > filesystems? If the question is in the context of WIN32, I suppose switching filesystems means that the paths will be absolute, and in that case, they would not be problematic, or...? -- You are receiving this mail because: You are on the CC list for the bug.
