Re: [Patch] FIrst experimental profiler

[Jordi Gutiérrez Hermoso]

2011/6/30 Daniel Kraft <address@hidden>:

> luckily, my holidays are drawing nearer -- and finally I have some
> more time as before already. Attached is my current working code for
> the first experimental profiler.

I'm sorry it's taken me so long to look into this. I've been out of
town with sparse internet access and just got home last night.

I tried to apply your patch, but it appears to be attached to a cset
that is not in Savannah. Do you have cset
fcdd244799c4ed9969dda351bb0bb79976252a09 ? If so, is *that* cset's
parent in Savannah? If not, can you provide the missing intermediate
cset or rebase your changes relative to Savannah? The easiest way to
rebase is with the mq extension (WARNING: this rewrites history, i.e.
potentially deletes data if you're not careful, which is why it's
disabled by default [1]):

    #Move your changes to a patch
    hg qimport -r $rev_number_of_your_cset -n temp_patch

    #Unapply the patch with the changes
    hg qpop

    #Get the latest changes
    hg pull $savannah_url
    hg up default

    #Reapply your patch, now based at the latest cset in Savannah
    hg qpush

    #Turn your patch back into an commit in history
    hg qfinish temp_patch

Your changes apply if I apply them as an ordinary patch foregoing
Mercurial, but I want to make sure that what you submitted is indeed
the changes you want before I start reviewing them. I don't want to
start the work only to later realise that you missed a small but
crucial detail in the cset that hg is missing.

Also, Google wants us to do midterm reports on our progress. We only
have two days, so we definitely should mention this patch and evaluate
your progress with it.

Thanks,
- Jordi G. H.

[1] I know this is a common and frequent occurrence in git with which
you are familiar, but hg makes a point of being safe by default, and
the community ethos is to not recommend unsafe practices to other
people without at least saying they're unsafe. It's a bit paranoid,
but it's better than losing data due to learners' mistakes.