Re: gcc-3.0.1 generates invalid assembler (undefined label) on sparc

[Jeff Sturm <jsturm at one-point dot com>]

On 1 Jan 2002, Niels Möller wrote:
> > I don't know what the equivalent on 3.0.3 is, perhaps remove_insn.
> 
> Are you suggesting that I try changing the delete_insn to remove_insn
> on that line? I did that, and it compiled, but testing if it works
> have to wait for tomorrow.

Yes, my hunch is that one of remove_insn or flow_delete_insn is more
appropriate there than delete_insn.  I really don't know that 3.0.x code
that well.

Failing that, you can use the workaround I posted earlier or try a 3.1
prerelease.  If you are stuck, submit a PR and someone will test it in the
3.1 timeframe.

> > You can enter a PR if you like, but I have a suspicion the eventual fix
> > will be "upgrade to 3.1" if that solves the problem.
> 
> When is that scheduled?

The timeline in http://gcc.gnu.org/develop.html shows the projected
release date for 3.1 is April 15 2002.

> I would have expected that bugs that generate
> invalid assembler be fixed on the 3.0.x branch, but I'm not very
> familiar with the gcc process.

Unfortunately there are still many bugs, as shown in GNATS.  My guess is
that only regressions are likely to be given priority in 3.0.x, and
development on the branch is likely to be slowed as volunteers turn their
attention to 3.1.  I don't recall when or if there'll be another 3.0.x
release.

I find that the current compiler (that will become 3.1) is quite stable
now, and use it as my daily compiler with few difficulties

Jeff