darcs

Patch 1532 Accept issue2275: darcs follows symbolic links instead...

Title Accept issue2275: darcs follows symbolic links instead...
Superseder Nosy List bf, gpiero
Related Issues
Status accepted Assigned To
Milestone

Created on 2017-03-08.11:37:39 by gpiero, last changed 2017-04-17.12:40:05 by ganesh.

Files
File name Status Uploaded Type Edit Remove
accept-issue2275_-darcs-follows-symbolic-links-instead-of-ignoring-them.dpatch gpiero, 2017-03-08.11:37:38 application/x-darcs-patch
patch-preview.txt gpiero, 2017-03-08.11:37:38 text/x-darcs-patch
See mailing list archives for discussion on individual patches.
Messages
msg19355 (view) Author: gpiero Date: 2017-03-08.11:37:38
1 patch for repository valentina:var/repos/darcs/screened:

patch 6047f02180470b5048e7128e979be8200bd153ac
Author: Gian Piero Carrubba <gpiero@rm-rf.it>
Date:   Wed Mar  8 14:25:48 CET 2017
  * Accept issue2275: darcs follows symbolic links instead of ignoring them
Attachments
msg19373 (view) Author: bf Date: 2017-03-15.18:22:12
I have screened this patch. I do think this needs a bit more discussion
wrt the intended behavior, see my remarks re issue2275, but I would like
to see this resolved.
msg19429 (view) Author: ganesh Date: 2017-04-06.04:41:49
bf, whats's next for this patch? I see there was some more discussion in 
issue2275, but I didn't read it in detail. Does this failing test specify 
behaviour we want?
msg19438 (view) Author: bf Date: 2017-04-08.10:12:03
The test captures parts of the requirement: the symbolic link should be
ignored and *not* resolved. But I maintain that Darcs should issue a
warning. There are two cases to consider:

(1) a symlink with a name that collides with a known file or directory
(known = in recorded or pending)

(2) a symlink that targets a known file or directory

In case (1) I think it would be wrong to "ignore the symlink" as that
would mean the file or directory is regarded as deleted. This would be
very surprising. One might even make this an error ("error: linkname has
been replaced with symbolic link to linktarget").

Case (2) is slightly more benign, but I would still want a warning
(something like "warning: ignoring symbolic link linkname -> linktarget").

Variants could include asking the user how to proceed (resolve, ignore,
terminate).
msg19448 (view) Author: ganesh Date: 2017-04-17.07:19:28
OK, so I'll accept this patch and the test case can be improved later to 
check for a future warning too.
History
Date User Action Args
2017-03-08 11:37:39gpierocreate
2017-03-15 09:07:52bfsetstatus: needs-screening -> needs-review
2017-03-15 18:22:13bfsetmessages: + msg19373
2017-04-06 04:41:50ganeshsetnosy: + bf
messages: + msg19429
2017-04-08 10:12:06bfsetmessages: + msg19438
2017-04-17 07:19:28ganeshsetmessages: + msg19448
2017-04-17 12:40:05ganeshsetstatus: needs-review -> accepted