darcs

Issue 1993 Darcs get fails, darcs repair lies

Title Darcs get fails, darcs repair lies
Priority bug Status given-up
Milestone Resolved in
Superseder Nosy List dmitry.kurochkin, simonpj
Assigned To
Topics

Created on 2010-11-15.23:30:23 by simonpj, last changed 2014-04-03.12:42:15 by bfrk.

Messages
msg13077 (view) Author: simonpj Date: 2010-11-15.23:30:22
This is darcs 2.4.4 on Windows. Several difficulties:

================
Repeated attempts to 
	darcs get --lazy http:: //darcs.haskell.org/ghc-new-co-16Sep10/ghc
failed, usually by hanging. Sometimes they started saying

Copying pristine 577 done, 61 queued. Bag.lhs.rej
Hash failure in http://darcs.haskell.org/ghc-new-co-16Sep10/ghc/_darcs/pristine.hashed/0000000537-c48de98288d4c1dffa0e70e7ef9dcfdef4871c475208df086ce1eddfc10cf344
Copying pristine 590 done, 48 queued. FiniteMap.lhs

and similar.

=================
The resulting repo seems to be ok despite these failures. I tried
	darcs pull
and got "No remote changes to pull in!"
I tried
	darcs repair
A long time later I got
....
Replaying patch 22585 done, 48 queued. Thu Sep  9 09:54:41 GMT Daylight Time 201
Replaying patch 22612 done, 21 queued. Wed Sep 15 11:57:12 GMT Daylight Time 201
Replaying patch 22632 done, 1 queued. Wed Sep 15 23:59:35 GMT Daylight Time 2010
The repository is already consistent, no changes made.

=============
But the repo is broken.  I trie
	darcs record
to get
darcs failed:  error opening _darcs\pristine.hashed\0000002011-21b431eca667eb0ace06ea663e851f21d8b617f56a3d29c835dfeff327d67b5d

=============

This is not good.    Why does darcs get fail?  Why does darcs repair claim all is well when it isn't?
 
(I did eventually succeed in getting darcs get --lazy to go through without error.)

Simon
msg13085 (view) Author: galbolle Date: 2010-11-17.12:38:51
do you have tarballs of the failing repository? (the one darcs get --lazy 
gave you and which you cannot record in).
msg13089 (view) Author: simonpj Date: 2010-11-17.16:04:09
Sadly not.  If you like I'll try to do more darcs gets to see if they fail, and if so whether the repo seems duff


| -----Original Message-----
| From: Florent Becker [mailto:bugs@darcs.net]
| Sent: 17 November 2010 12:39
| To: dmitry.kurochkin@gmail.com; florent.becker@ens-lyon.org; Simon Peyton-Jones
| Subject: [issue1993] Darcs get fails, darcs repair lies
| 
| 
| Florent Becker <florent.becker@ens-lyon.org> added the comment:
| 
| do you have tarballs of the failing repository? (the one darcs get --lazy
| gave you and which you cannot record in).
| 
| ----------
| status: unknown -> waiting-for
| 
| __________________________________
| Darcs bug tracker <bugs@darcs.net>
| <http://bugs.darcs.net/issue1993>
| __________________________________
msg13153 (view) Author: mornfall Date: 2010-11-21.14:59:05
Simon, can you please share darcs +RTS --info? We are seeing network-
related hangs in darcs built with GHC 6.10 here. It's just a wild guess, 
but it's worth checking...
msg13203 (view) Author: kowey Date: 2010-11-22.10:50:38
If I may, Simon recently submitted three bugs that seem to have 
something in common about not being able to fetch files (or something 
going wrong when doing so):
- issue1969
- issue1993
- issue2000

Perhaps they're all manifestations of the same underlying problem?
msg13214 (view) Author: simonpj Date: 2010-11-22.12:31:03
Volia!

darcs +RTS --info
 [("GHC RTS", "YES")
 ,("GHC version", "6.10.4")
 ,("RTS way", "rts_thr")
 ,("Host platform", "i386-unknown-mingw32")
 ,("Host architecture", "i386")
 ,("Host OS", "mingw32")
 ,("Host vendor", "unknown")
 ,("Build platform", "i386-unknown-mingw32")
 ,("Build architecture", "i386")
 ,("Build OS", "mingw32")
 ,("Build vendor", "unknown")
 ,("Target platform", "i386-unknown-mingw32")
 ,("Target architecture", "i386")
 ,("Target OS", "mingw32")
 ,("Target vendor", "unknown")
 ,("Word size", "32")
 ,("Compiler unregisterised", "NO")
 ,("Tables next to code", "YES")
 ]
bash-3.1$

|  -----Original Message-----
|  From: Petr Ročkai [mailto:bugs@darcs.net]
|  Sent: 21 November 2010 14:59
|  To: dmitry.kurochkin@gmail.com; me@mornfall.net; Simon Peyton-Jones
|  Subject: [issue1993] Darcs get fails, darcs repair lies
|  
|  
|  Petr Ročkai <me@mornfall.net> added the comment:
|  
|  Simon, can you please share darcs +RTS --info? We are seeing network-
|  related hangs in darcs built with GHC 6.10 here. It's just a wild guess,
|  but it's worth checking...
|  
|  __________________________________
|  Darcs bug tracker <bugs@darcs.net>
|  <http://bugs.darcs.net/issue1993>
|  __________________________________
msg13222 (view) Author: mornfall Date: 2010-11-22.19:39:30
Hello,

Simon Peyton-Jones <simonpj@microsoft.com> writes:
> darcs +RTS --info
>  [("GHC RTS", "YES")
>  ,("GHC version", "6.10.4")
>  ,("RTS way", "rts_thr")
>  ,("Host platform", "i386-unknown-mingw32")
>  ,("Host architecture", "i386")
>  ,("Host OS", "mingw32")
>  ,("Host vendor", "unknown")
>  ,("Build platform", "i386-unknown-mingw32")
>  ,("Build architecture", "i386")
>  ,("Build OS", "mingw32")
>  ,("Build vendor", "unknown")
>  ,("Target platform", "i386-unknown-mingw32")
>  ,("Target architecture", "i386")
>  ,("Target OS", "mingw32")
>  ,("Target vendor", "unknown")
>  ,("Word size", "32")
>  ,("Compiler unregisterised", "NO")
>  ,("Tables next to code", "YES")
>  ]

now the next question is: do you think you could compile a darcs binary
with 6.12? (Unfortunately, we don't quite work on GHC 7 yet... and
congrats on the release, by the way...)

I guess it's a bit tricky with cURL on windows, but you could try with
-f-curl: it may be a bit slower on HTTP downloads, although I guess slow
but working is better than fast but hanging half the time.

I am not sure whether I should recommend going with 2.5 at the occasion:
it ought to be faster on some operations (especially on the hashed GHC
repo), although it may also come with new and interesting bugs (I have
my share of guilt in that, admittedly). On the other hand, many of us
have been using it for a fairly long time without running into anything
serious, as far as I can tell. It's ultimately your call. :)

Either way, please let us know how it is going. If you can't build for
whatever reason, we can look around for a better Windows binary for
you...

Yours,
   Petr.
msg13223 (view) Author: mornfall Date: 2010-11-22.19:44:41
Hi again,

Petr Ročkai <bugs@darcs.net> writes:
> now the next question is: do you think you could compile a darcs binary
> with 6.12? (Unfortunately, we don't quite work on GHC 7 yet... and
> congrats on the release, by the way...)

unfortunately, 6.10 might have been a red herring. I have just noticed a
hang on the 6.12 windows build as well. Sorry about that. I'll keep you
informed if I have anything new.

Yours,
   Petr.
msg13229 (view) Author: simonpj Date: 2010-11-23.16:26:54
| now the next question is: do you think you could compile a darcs binary
| with 6.12? (Unfortunately, we don't quite work on GHC 7 yet... and
| congrats on the release, by the way...)

| I guess it's a bit tricky with cURL on windows, but you could try with
| -f-curl: it may be a bit slower on HTTP downloads, although I guess slow
| but working is better than fast but hanging half the time.

I really, really don't want to get into building Darcs from source. I'm just a naive darcs user reporting a bug.

Simon
msg17255 (view) Author: bfrk Date: 2014-04-03.12:42:13
I don't think this will ever get resolved, considering that ghc HQ moved
to git.
History
Date User Action Args
2010-11-15 23:30:23simonpjcreate
2010-11-17 12:38:51galbollesetstatus: unknown -> waiting-for
messages: + msg13085
2010-11-17 16:04:10simonpjsetmessages: + msg13089
2010-11-21 14:59:06mornfallsetmessages: + msg13153
2010-11-22 10:50:39koweysetmessages: + msg13203
2010-11-22 12:31:04simonpjsetmessages: + msg13214
2010-11-22 19:39:30mornfallsetmessages: + msg13222
2010-11-22 19:44:42mornfallsetmessages: + msg13223
2010-11-23 16:26:55simonpjsetmessages: + msg13229
2014-04-03 12:42:15bfrksetpriority: bug
status: waiting-for -> given-up
messages: + msg17255