I built a darcs-unstable from today's sources (18 Feb 2008), and it hangs 
eating all my memory whan I try to 'darcs check' a GHC repository (darcs2 
format).

The reason I'm trying this is because I have a darcs2 from a couple of 
weeks ago built with bytestring support, that fails to 'darcs check' the 
GHC repository, claiming an inconsistency.  I'll report this separately, as 
the issue may still exist.

Simon,

Are there any particularly huge patches in this repo? Here's a hackish way to
check for patches bigger than 1 meg:

du -hs _darcs/patches/* | grep M

Also, Does "--debug-verbose" provide any insight into the cause of this?

( I assume that flag exists for "check"... I don't have a darcs-2 binary handy). 


 Mark

Mark Stosberg wrote:

> Are there any particularly huge patches in this repo? Here's a hackish way to
> check for patches bigger than 1 meg:
> 
> du -hs _darcs/patches/* | grep M
> 
> Also, Does "--debug-verbose" provide any insight into the cause of this?
> 
> ( I assume that flag exists for "check"... I don't have a darcs-2 binary handy). 

It's not just 'darcs check', just doing 'darcs changes' takes for ever on 
this repo now.  I've just tried it with --debug-verbose, and it's printing

I'm doing copyFileUsingCache on 
patches/1c5ece7eed6f8ccf40f8a96789ccb74fbbdd29de
I'm doing copyFileUsingCache on 
patches/dad1a6a5afc02c6310c08731feff0d2336d18f75
....

but getting slower over time, and the darcs process is up to 430MB or so 
(but not growing any more, it seems).

To answer your question, here are the largest patches in the repo:

$ ls --sort=size -l | head
total 104236
-rw-rw-r-- 2 simonmar GHC 3413053 Dec 11 17:18 
74f446e288b9f13dadd687eac3ee57aaf052bc5d
-rw-rw-r-- 2 simonmar GHC 3162732 Dec 11 20:41 
476ee36154bc7d46ed1bdfad5b5fb193ac25bb3f
-rw-rw-r-- 2 simonmar GHC 1772377 Dec 11 18:15 
020364f7429a56fcfc0ccd238737c6bef8808f7b
-rw-rw-r-- 2 simonmar GHC  873365 Dec 11 17:31 
902e2c518a21cfc95935bfe4215cd0d0732d100d
-rw-rw-r-- 2 simonmar GHC  817974 Dec 11 17:42 
fa0c1ccce2251240f7f19273470b653534c97a52
-rw-rw-r-- 2 simonmar GHC  734741 Dec 11 19:27 
1a37c2a93d700fa5f18b6e30443b60b649d5d771
-rw-rw-r-- 2 simonmar GHC  539293 Dec 11 18:00 
6a9201dfc233d4c86b004b1f0dcc5c39476995c0
-rw-rw-r-- 2 simonmar GHC  526646 Dec 11 17:21 
031ffe22d4d58f8e31a84bdbfde19c0e793ccbf4
-rw-rw-r-- 2 simonmar GHC  523619 Dec 11 20:01 
b65dc151e979931f36601a8fb37553b7f84bb27e

Cheers,
	Simon

I'm adding this to the "release critical" list, at least until we can determine
it's not a regression.

I've reproduced this on a hashed copy of the ghc repository, and can confirm
that it's definitely a serious bug.  (But this is as far as I've gotten, so far.)

The following patch updated the status of issue699 to be resolved in the unstable branch:

* resolved issue699: revert accidentally-increased strictness in repo-reading. 

You can view the patch details online here: 
http://darcs.net/cgi-bin/darcs.cgi/unstable/?c=annotate&p=20080222173129-72aca-09d70caeea9682fee61415319b18b5964512b6a4.gz