one byte change to file not detected ?

[Adrian Buehlmann]

On 20.02.2008 22:29, Bryan O'Sullivan wrote:
> BuraphaLinux Server wrote:
> 
>> I am writing a low-tech, brute-force RCS-->hg converter.  It plods
>> through each file, pulling each successive rcs version and then doing
>> an hg commit.
> 
> I'd suggest writing something within the existing convert framework
> instead.  It will be much faster and safer than what you're currently doing.

I made a somewhat similar experience as Burapha a couple of weeks ago (but with 
a slightly different outcome :-)

I've identified my problem and documented it at the bottom of
http://www.selenic.com/mercurial/wiki/index.cgi/DirState

My (newbie) solution was to wait until the next second toggles
(did that in my Python beginner's piece I documented at
http://www.selenic.com/mercurial/wiki/index.cgi/p4hg).

I then started to think about why Mercurial does it the way it does it and 
became a fan of Mercurial when I realized *why* it does it like this.

With Perforce, I had to tell Perforce which file I want to "edit". I realized 
how stupid that actually is and how often I "p4 edited" a whole tree and "p4 
reverted" the unchanged files.

I wouldn't want to go back to Perforce. Mercurial's approach is better, but you 
have to understand it and be careful with serial scripted hg commit calls - the 
classical Mercurial newbie trap.

Hmm. Maybe I should add a note about that on
http://www.selenic.com/mercurial/wiki/index.cgi/Commit.

See also http://www.selenic.com/mercurial/bts/issue618