Re: how close are we to having an "addAt" or "insertAt" feature?

[David Kastrup]

Werner LEMBERG <address@hidden> writes:

>>> And the wortliste still takes more than 6 minutes on my computer:
>>
>> Which is not even a factor of 3 as compared to previously:
>
> But even that is an large improvement.  It's now unbearable and no
> longer unusable :-)

So since much of the load appears to be I/O bound, I did "gc
--aggressive" on the Emacs repository I had been testing with, reducing
repository size to about a quarter.  The performance of "git blame" has
become considerably worse by that.  I suppose that the time is dominated
by unpacking the repository now: half of the run time is "system", and I
am using an SSD disk.  For rotary disks, the actual improvements will
likely be even less.

So it would seem that for further improvements to git-blame, I'll have
to work on the core git mechanisms and/or let git-blame do smart caching
to avoid redundant data accesses.

Very disappointing.  I don't have the time to rewrite the core of git.

-- 
David Kastrup