| Summary: | emerge gnat-3.43 fails complaining about inconsistent version of gcc/gnatbind | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Marek Janukowicz <marek> |
| Component: | [OLD] Development | Assignee: | ada team [OBSOLETE] <ada+disabled> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | darktjm |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | x86 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | Possible fix | ||
|
Description
Marek Janukowicz
2005-03-24 01:17:09 UTC
Created attachment 57457 [details, diff]
Possible fix
This ebuild does not play well with a preexisting gnat install, apparently
using both it and the bootstrap compiler. To fix, set ADA_INCLUDE_PATH and
ADA_OBJECTS_PATH. The attached a preliminary patch does this, as well as
possibly fixing this build for powerpc, and adding a flag to allow using any
installed gnat as the bootstrap compiler. I haven't fully tested either the
noboot flag or powerpc yet, though, since my powerpc is disassembled and I
didn't add the noboot until 3.4.3 was installed, so it might not work with
others. Feel free to remove the noboot stuff or rename the use flag, or
document it in the right place, or whatever. Note that there is probably a
better way to do this, but I only know gnat well enough to get it to compile my
own programs.
The patch fixed the issue for me, thanks So, who closed the bug while the fix did not make it into the tree yet?
Marek: if that was you, please do not do so in the future. This effectively
makes the bug loose visibility and we are much less likely to react on it..
Thomas: Thanks for the patch and for testing and commenting. I incorporated
what was not yet implemented of this patch in the 3.44 version. This really
boiled down to defining ADA_OBJECTS_PATH and ADA_INCLUDE_PATH. The rest of
assignments were already done. I cleaned-up detection of GCC_EXEC_PREFIX -
should now work on all three "supported" arches, so this should not cause that
failure on x86 you reported to me yesterday..
The noboot flag and related branches I nuked for now.
David: I would rather leave this at your discretion, I mean whether to
implement this. The idea of noboot flag, as I understand from the patch, is to
allow building gnat using installed compiler. IMHO this does not give any real
advantages (the compiler is bootstrapped and built against final internal
version anyway) and can potentially cause problems such as the reported one.
The fixes are in the tree, I replaced previous version of gnat-3.44.ebuild. No
revision bump, as these are compile-time fixes and do not provide any new
functionality.. well, the usual drill. I'll leave the bug closed, since the fix
is in, but if anybody tests and finds a problem, please reopen..
The shared libs are still commented out (#64373) and the ebuild is still
masked. Waiting for more tests/comments..
George
|
