Re: 1894 GCC regressions, 1878 new, with your patch on2002-07-24T18:30:44Z.

[cgd at broadcom dot com]

At Wed, 24 Jul 2002 16:20:04 -0700, Geoff Keating wrote:
> The problem with that is that the tester won't update its binutils or
> newlib unless the tree has no regressions, because otherwise it can't
> tell if it's updating to a broken binutils or newlib.  In addition, it
> can't upgrade right now anyway because gdb introduces new regressions.

Hmm.  Chicken and egg.


> So, can you add some backwards-compatibility for old newlibs?

Any suggestions for this?

the only thing i can think of is adding a test to gcc/configure.in
which checks whether a specific mips libgloss linker script that has
existed "forever" (e.g. lsi.ld) exists in the source tree and includes
a STARTUP file.  Then, use that test to #ifdef in the gcc/config/mips
hdrs.

Does that seem reasonable?

(it's ... kinda gross.  will only work for combined-source-tree
builds, but i can't think of a better thing to try...  then, how long
does it have to live, since it will be fairly icky...)



chris