This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

[new-regalloc-branch] No patch


Beware, this is a rant.

Hi,

I just saw, that something was committed to this branch without any
indication on this list (at least not in the archives).  This is
frustrating:  I took the last two days and nights to fix the branch.  It
wasn't even bootstrapping, produced many regressions, and there were
obvious errors in the last patches. SPEC2000 still has problems, even
after multiple hours of debugging, but at least these errors are also
happening with -fno-new-ra.  Unfortunately there are also other rather
large changes to modules not connected to the register allocator, so I'm
not sure, if this is also happening with the mainline at the merge-point,
or can be attributed to these unrelated changes.  The only thing I know
is, that at some point in the past (specifically before the latest
mainline merge, but after my 10/10 diff last month) SPECINT2000 was
completely working (without eon).  Whatever.  I'm now not going to again
test and fix the changes from now, at least not today, I need sleep.

Instead I have good faith that these latest commits also fixed these
errors. If I'm tomorrow checking out this branch, and it's not right out
of the box bootstrapping on x86, or produces more regressions than before
I'll be very annoyed.  I will be less annoyed, if that checkout fixes also
regressions that I locally fixed the last days.


Ciao,
Michael.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]