Re: [Radmind-devel] [ radmind-Patches-2942523 ] Fix for replacing dirs with symlinks

[Jeff K. <sla...@um...>]

I've attached a new version of this patch to the original. It defines a new status, PR_REPLACE, that does everything PR_DOWNLOAD does except that it doesn't add the leading - to the to-be-printed line but does set fs_minus.

Jeff Kelley
sla...@um...
Campus Computing Sites
Information and Technology Services
University of Michigan

On Feb 18, 2010, at 11:06 AM, Jeff Kelley wrote:

> Andrew,
> 
> 	I've updated the patch description with an example of how to reproduce the issue it hopes to fix. As far as re-doing the patch is concerned, it looks like this is the relevant portion of t_print that handles this for directories to be deleted:
> 
>    if ( print_minus ) {
>    /* set fs_minus so we can handle excluded files in dirs to be deleted */
>    fs_minus = 1;
>    fprintf( outtran, "- " );
>    }
> 
> 	We don't want to set print_minus in this case, since the first line will not be a deletion (it'll be the link). If you want to avoid inspecting fs and tran in t_print, then it looks like we'll need a flag other than PR_DOWNLOAD to pass to t_print. Maybe PR_REPLACE?
> 
> Jeff Kelley
> sla...@um...
> Campus Computing Sites
> Information and Technology Services
> University of Michigan
> 
> On Feb 9, 2010, at 11:45 AM, Andrew Mortensen wrote:
> 
>> I did see it. Don't set fs_minus in t_print: that routine is for output, not changing state.
>> 
>> Please also update your report with the information you've got here, and include minimal command file + transcripts exercising the bug.
>> 
>> andrew
>> 
>> On Feb 8, 2010, at 10:49 PM, Jeff Kelley wrote:
>> 
>>> I just wanted to make sure people saw this patch I sent in. I've confirmed it works for cases where you have the following:
>>> 
>>> 	/some/dir
>>> 	/some/dir/foo
>>> 
>>> where 'foo' matches an exclude pattern. Replacing the loadset with this:
>>> 
>>> 	l /some/dir -> /some/other/dir
>>> 
>>> results in an error removing the directory /some/dir, as it isn't empty. The patch sets 'fs_minus' to 1 in cases where the transcript has a link and the filesystem has a folder. With this patch, fsdiff -A generates a line that removes /some/dir/foo, resolving the problem.
>>> 
>>> Please let me know if there's anything else you need me to do for this patch. Also, see bug 1985781, which I submitted a while back but couldn't reproduce. This patch will fix that issue. Thanks in advance for your help.
>>> 
>>> Jeff Kelley
>>> sla...@um...
>>> Campus Computing Sites
>>> Information and Technology Services
>>> University of Michigan
>>> 
>>> On Jan 29, 2010, at 4:34 PM, SourceForge.net wrote:
>>> 
>>>> Patches item #2942523, was opened at 2010-01-29 16:34
>>>> Message generated for change (Tracker Item Submitted) made by slaunchaman
>>>> You can respond by visiting: 
>>>> https://sourceforge.net/tracker/?func=detail&atid=749494&aid=2942523&group_id=141444
>>>> 
>>>> Please note that this message will contain a full copy of the comment thread,
>>>> including the initial issue submission, for this request,
>>>> not just the latest update.
>>>> Category: UNIX
>>>> Group: v1.X
>>>> Status: Open
>>>> Resolution: None
>>>> Priority: 5
>>>> Private: No
>>>> Submitted By: Jeff Kelley (slaunchaman)
>>>> Assigned to: Nobody/Anonymous (nobody)
>>>> Summary: Fix for replacing dirs with symlinks
>>>> 
>>>> Initial Comment:
>>>> If you have a directory that is to be replaced with a symlink, and that directory has an excluded item in it, fsdiff doesn't create a line for lapply to remove the file. This patch fixes that behavior.
>>>> 
>>>> ----------------------------------------------------------------------
>>>> 
>>>> You can respond by visiting: 
>>>> https://sourceforge.net/tracker/?func=detail&atid=749494&aid=2942523&group_id=141444