diff problem

[Tim Armstrong]

Oops.. sorry for the omission of system information and version numbers.
I have the problem on Windows only. I am running Windows XP SP3 on a 
machine with 2.5 GB of RAM. I was able to reproduce the problem running 
Mercurial 1.3.1 as well as Mercurial 1.4.1.
Ubuntu in a virtual machine on the same box does not exhibit this 
problem, whether with the Ubuntu shipped version 1.1.2 or with version 
1.4.1 which I built from source.
I am not quite savvy enough to build mercurial from source on Windows, 
so I was unable to check if the patch supplied by Alistair fixes the 
problem.
Thank you very much for your time,
Tim

Matt Mackall wrote:
> On Tue, 2009-12-08 at 16:42 +0700, Tim Armstrong wrote:
>   
>> Hi,
>> I recently attempted to commit a 5.3 MB file with 5.7 MB worth of 
>> changes to its most recent revision (which was 5.1 MB in size). The 
>> commit failed and as I further investigated the cause I discovered that 
>> hg diff also fails after about 40 seconds by simply exiting without 
>> giving an error message. If I simply delete half of the new file for 
>> testing purposes the diff succeeds. The exact file size at which it 
>> fails seems to vary though so it doesn't seem like a problem with the 
>> file content.
>> I found this message on the mercurial list 
>> http://marc.info/?l=mercurial&m=117499200905031&w=2 but am unsure 
>> whether lack of memory if the cause of my problem as the file sizes do 
>> not seem excessive (especially compared to the sizes that other folks on 
>> the list mention having problems with) and I am able to get a diff from 
>> the diff program as well as kdiff3 and meld.
>> I would appreciate any thoughts or insight,
>>     
>
> At an absolute minimum, basically any bug report should include a
> version number. But we'll probably also want to know other relevant
> information like what operating system you're using, how much memory you
> have, and whether any resource limits are in place (on Unix/Linux with
> bash, that'd be ulimit -a).
>
> You might also try running your command with --debug --traceback.
>
>