[issue1286] hg normalization performance on Win32 (case folding systems)

[Petr Kodl]

New submission from Petr Kodl <petrkodl at gmail.com>:

under some circumstances the path normalization causes very slow repo operations 

example: hg init in large repository immediately followed by hg stat 

the problem is that foldmaps are empty for freshly initialized repositories so
the normalize call in dirstat that is normally used for each file in the repo 
will always call the util.fspath for each file 

given that the work list is result of path scanning the normalization and
additional lookups should not be required 

proposed patch is attached  
on ~30k files this accelerates the hg init/hg stat sequence from 20s to .56s

----------
files: no_normalization.txt
messages: 6932
nosy: pko
priority: wish
status: unread
title: hg normalization performance on Win32 (case folding systems)

____________________________________________________
Mercurial issue tracker <mercurial-bugs at selenic.com>
<http://www.selenic.com/mercurial/bts/issue1286>
____________________________________________________
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: no_normalization.txt
Url: http://selenic.com/pipermail/mercurial-devel/attachments/20080903/b2595d96/attachment.txt