|
Created on 2008-09-10.14:36:39 by simonpj, last changed 2011-07-05.09:59:28 by mornfall.
msg5975 (view) |
Author: simonpj |
Date: 2008-09-10.14:36:37 |
|
Should this happen?
bash$ darcs annotate compiler/basicTypes/DataCon.lhs
Stack space overflow: current size 8388608 bytes.
Use `+RTS -Ksize' to increase it.
8M is quite a big stack.
It worked with 100M (albeit very slowly).
Simon
|
msg5977 (view) |
Author: kowey |
Date: 2008-09-10.16:06:46 |
|
Simon: we should certainly avoid this happening if we can help it!
In the meantime, you can use
darcs changes -v --only-to-file compiler/basicTypes/DataCon.lhs
as a crude workaround.
Darcs people: I have reproduced this, including the +RTS -K100M workaround.
See also http://bugs.darcs.net/issue1017 for a stack overflow when doing
whatsnew on a large file.
I need a volunteer to boil this down into a minimal test case, if possible. The
minimal test case may not actually be very practical for the darcs test suite,
because it may involve generating lots of patches to a single file. Volunteer
should have a look at maybench (darcs get http://code.haskell.org/maybench )
because there is a darcs-benchmark tool which creates repositories automatically.
|
msg6485 (view) |
Author: beschmi |
Date: 2008-10-27.19:51:22 |
|
I'm working on optimizing annotate and changes, which should make both
faster and also fix this bug.
Benedikt
|
msg7804 (view) |
Author: twb |
Date: 2009-05-13.02:25:13 |
|
Hi Benedikt,
On Mon, Oct 27, 2008 at 09:51:22AM +0000, Benedikt Schmidt wrote:
> I'm working on optimizing annotate and changes, which should make
> both faster and also fix this bug.
Is this work now done? Can we close this bug?
|
msg13369 (view) |
Author: simon |
Date: 2010-12-16.18:29:29 |
|
Update: no, this bug is still very much with us:
~/src/ghc$ time darcs annotate README
Stack space overflow: current size 8388608 bytes.
Use `+RTS -Ksize -RTS' to increase it.
real 4m49.721s
user 1m14.573s
sys 0m16.529s
This issue is marked "superseded by #1566", which looks as if it might
be not far from landing. If that's not the case, an interim usability
fix for this would be good - it's an easily encountered stack overflow
that was reported two years ago and is still scaring new users away from
darcs (heard from kmc on #haskell today). Is it feasible to guestimate
the size of the annotate operation ahead of time and exit with a
friendly warning instead of a run-time error ? If not, even printing a
warning about this issue before attempting any annotate would cushion
the blow.
|
msg13371 (view) |
Author: simon |
Date: 2010-12-16.19:26:52 |
|
Patch "annotate interim usability fix: print a warning about issue #1068"
sent to patches@darcs.net.
|
msg14566 (view) |
Author: mornfall |
Date: 2011-07-05.09:59:27 |
|
Shouldn't happen with the new annotate anymore. Benedikt's improvements
would still be useful, although not strictly necessary to fix this
particular bug.
|
|
Date |
User |
Action |
Args |
2008-09-10 14:36:39 | simonpj | create | |
2008-09-10 16:06:50 | kowey | set | status: unread -> unknown priority: bug topic:
+ Performance messages:
+ msg5977 title: Stack overflow -> annotate stack overflow |
2008-09-10 16:06:59 | kowey | set | status: unknown -> needs-reproduction |
2008-10-12 19:28:24 | thorkilnaur | link | issue1145 superseder |
2008-10-27 19:51:24 | beschmi | set | status: needs-reproduction -> has-patch nosy:
+ dmitry.kurochkin, simon, beschmi, thorkilnaur messages:
+ msg6485 assignedto: beschmi |
2009-05-13 02:25:16 | twb | set | nosy:
+ twb messages:
+ msg7804 |
2009-08-06 21:12:28 | admin | set | nosy:
- beschmi |
2009-08-10 23:45:16 | admin | set | nosy:
- dagit |
2009-08-25 17:32:47 | admin | set | nosy:
+ darcs-devel, - simon |
2009-08-26 13:47:41 | kowey | set | status: has-patch -> deferred nosy:
kowey, darcs-devel, simonpj, twb, thorkilnaur, dmitry.kurochkin superseder:
+ patch index optimisation (aka filecache) |
2009-08-27 14:33:07 | admin | set | nosy:
kowey, darcs-devel, simonpj, twb, thorkilnaur, dmitry.kurochkin |
2010-12-16 18:29:30 | simon | set | status: deferred -> unknown messages:
+ msg13369 |
2010-12-16 19:26:52 | simon | set | messages:
+ msg13371 |
2011-07-05 09:59:28 | mornfall | set | status: unknown -> resolved messages:
+ msg14566 resolvedin: 2.8.0 |
|