Re: [Gnu-arch-users] why does "tla get" with many patches become slower and slower?

[Tom Lord]

    > From: Denys Duchier <address@hidden>

    > Do you really think it is reasonable to require almost 1.5 hours
    > to apply 747 patches?

That would depend on the size of your tree and the patches.  12s for a
whole-tree patch is not necessarily excessive -- configuring things so
that you need to apply 747 patches on a regular basis may suggest a
user error.

But mostly that's beside the point.  Didn't you initially report
seeing the average _time_per_patch_ climb sharply as hundreds of
patches were applied?  (Or am I confusing you with someone else?)

Assuming that the tree-size isn't growing by leaps and bounds and that
the changesets themselves are all in the same ballpark, the time per
patch in a long run like that should be at least close to flat.  If
it's not, that suggests the possibility of a bug and the way to start
looking for that is with some memory and CPU profiling.

-t