Re: xdelta, svndiff, zlib, or some other problem

Edward Ned Harvey <svn_at_nedharvey.com> writes:

> So then I repeated the above process, using one of the problem scenarios. A
> commit which takes 15 minutes and should take 11 mins. (For some reason,
> this time it only took 8 min.) But as I read the output, I see numbers like
> "40" calls to something ... which don't sound like 8 mins worth of
> processing.
>
> So I'm not sure I did this right.
>
> The output was:
>
> Flat profile:
>
> Each sample counts as 0.01 seconds.
> no time accumulated
>
> % cumulative self self total
> time seconds seconds calls Ts/call Ts/call name
> 0.00 0.00 0.00 42 0.00 0.00 data_start
> 0.00 0.00 0.00 1 0.00 0.00 svn_cl__args_to_target_array_print_reserved

You have done something wrong, very little data has been recorded.
It's possible you didn't rebuild the libraries, or you are picking up
the wrong libraries at runtime.

On Linux I would use oprofile rather than gprof.

-- 
Philip