[PATCH 0 of 1 STABLE] revset: evaluate sub expressions correctly (issue3775)

Matt Mackall mpm at selenic.com
Tue Jan 22 18:01:57 CST 2013


On Wed, 2013-01-23 at 00:26 +0900, FUJIWARA Katsunori wrote:
> This patch series fixes issue3775.
> 
> BTW, some revset implementations use "list(repo)" as "entire revisions
> in the repo", and it seems to be inefficient, because it causes
> immediate creation of list object containing all revision numbers in
> the repo.
> 
> IMHO, "repo.changelog" should be used instead of it.

Maybe. See

a) past discussion with bos about range objects for revsets
b) filtered changelog work from Pierre-Yves

> But it also seems to have not so serious impact for average revset
> expressions (not used repeatedly in the loop, for example). So I'm
> planing to send the patch to fix it after current code freeze.
> 
> Or should I send it in this code freeze, too ?

Performance issues that are not _new in this release_ are probably not
candidates for stable.

Functionality issues might be.

-- 
Mathematics is the supreme nostalgia of our time.




More information about the Mercurial-devel mailing list