[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

RE: inconsistency between mergeinfo records

From: Bert Huijben <bert_at_qqmail.nl>
Date: Tue, 24 Feb 2015 18:27:07 +0100

I see a few questions, that our merge experts over here on the dev@ list might have a better answer for than I have.




From: Stefan Hett [mailto:stefan_at_egosoft.com]
Sent: dinsdag 24 februari 2015 11:28
To: Bert Huijben; 'subversion'
Subject: Re: inconsistency between mergeinfo records


Hi Bert,

thanks. That mostly does explain the current behavior to me.
From a user's point of view I however find this difference in recorded mergeinfos quite problematic. While certainly both cases represent the same logical merge structure:
case 1:
svn:mergeinfo for /B: /A:2-5
case 2:
svn:mergeinfo for /B: /A:2-5
svn:mergeinfo for /B/test.txt /A/test.txt:3

The redundant? mergeinfo of /B/test.txt is causing quite some issues for us. It's true, that when I directly reintegrate B back into A, A would not record the "redundant" mergeinfo for A/test.txt. But if I create another branch from B (let's say C) and reintegrate this back into A, the mergeinfo (assuming, didn't test!) will be kept in /A/test.txt - ending up with mergeinfos on a file.
When I never reintegrate B back into A, this mergeinfo in test.txt wil stay forever, having an accumulating effect on the number of files containing mergeinfos over the time...

In our productive environment this now resulted in hundreds of files having retrieved this kind of redundant mergeinfos:
/XRebirth/branches/XR/data/shaderfx/high/XU_ALPHA8_LIT.FBPC:184223,184357,184562,184564,184569,184575,184642,184656,184658,184664,184666,184668,184670,184676,184690,184692,184703,184706,184714,184718,184724,184726,184742,184748,184752,184754,184758,184765,184768,184770,184772,184795,184797,184805,184808,184819,184837-184838,184849-184850,184890,184906,184942,184944,184965,184969,184987,185001,185045,185051,185053,185064,185071,185073,185075,185088,185093,185096,185111,185120,185142,185148,185154,185161,185182,185231,185270,185273,185301,185330,185332,185348,185357,185372,185374,185406,185426,185455,185461,185511,185526,185546,185559,185562,185566,185579,185606,185645,185669,185672,185674,185676,185678,185680,185689,185704,185738,185745,185749,185758,185797,185890,185893,185896,185898,185900,185909,185949,185993,186001,186007,186020,186031,186080,186082,186106,186108,186122-186123,186127,186134-186137,186166,186169,186172,186174,186181,186183,186186,186210,186214,186218,186225,186234 ,186239,186248-186249,186259,186265,186269,186272,186286,186290,186302,186318,186334,186344,186357,186360-186361,186380,186382,186405,186420,186447,186456-186458,186466,186471,186506,186511,186543,186561,186566,186583-186584,186605,186607,186609,186614,186616,186620,186623,186635,186644,186646,186661,186665,186668,186673,186683,186685,186693,186700,186702,186706,186714,186717,186727,188312,190701-190708,190953-190954,190967,191011,191021,191055,191057,191062,191104,191110,191113,191125,191171,191181,191183,191185,191238,191249,191251,191253,191260,191302,191324,191326,191352,191366-191367,191407-191408,191412,191429,191471,191494,191513,191524,191532,191537,191540,191554,191606,191636,191656,191660,191675,191695,191701,191706,191709,191712,191714,191735,191740-191741,191782,191794,191809,191812,191834,191846,191856,191860,191882

Using TortoiseSVN as our main client, this makes a lot of cases quite inconvenient:
- showing the overview when committing merged changes, is hard, because this brings up a list of hundreds of files with the actual changed files being somewhere in-between
- logs are cluttered with mergeinfo changes, so it's quite hard to find the actual changes in a revision
- committing changes is unnecessarily slowed-down because all mergeinfo changes are transferred one-by-one
- I guess other SVN-operations are slowed-down as well, because the mergeinfos are not stored only in one single mergeinfo-property...

Do u have any suggestion for us to improve our workflow? Wouldn't it be reasonable to change the behavior of the --record-only merge, so that it does not store these "redundant" mergeinfos in the first place?


I haven’t looked at the full details, but actual merges only store mergeinfo of what is actually merged (includes unaffected tree filtering, filtering what is already merged, etc.). A record only merge is a tool to just register as merged the affected target without further processing. It is primarily used to block further merges, or unblock something that wasn’t really merged.


So totally different mergeinfo is fully expected.


Does this answer your question, or did either of your operations record wrong mergeinfo?




Sent from Windows Mail


From: Stefan Hett <mailto:stefan_at_egosoft.com>
Sent: ‎Monday‎, ‎February‎ ‎23‎, ‎2015 ‎8‎:‎30‎ ‎AM
To: 'subversion' <mailto:users_at_subversion.apache.org>


Another user (Sergey Azarkevich) actually pointed me to an interesting fact:
C:\test\test2checkout>svn merge file:///C:/test/test2/A <file:///C:\test\test2\A> B --record-only
--- Recording mergeinfo for merge of r2 through r5 into 'B':

C:\test\test2checkout>svn merge file:///C:/test/test2/A <file:///C:\test\test2\A> B
--- Merging r3 through r5 into 'B':

Using explicit range of revisions same as for --record-only lead to equal
modifications in wc:
C:\test\test2checkout>svn merge file:///C:/test/test2/A <file:///C:\test\test2\A> B -r 2:5
--- Merging r3 through r5 into 'B':

Note the different ranges (r2-r5 vs. r3-r5 in the first two calls).
Maybe this sheds some light here?

> Looks like the batch-file got truncated by some clients/mail servers
> on the way --- here's the plain batch file content.
> Anyone having an idea what's going on here?
> REM create test repository
> mkdir C:\test
> cd /d C:\test
> mkdir test2
> svnadmin create test2
> REM check-out test repository
> mkdir test2checkout
> svn co file:///C:/test/test2 <file:///C:\test\test2> ./test2checkout
> cd test2checkout
> REM add initial structure
> mkdir A
> echo > A\test.txt
> svn add A
> svn commit -m test
> REM copy A to B
> svn cp A B
> svn commit -m test
> REM modify A/test.txt
> echo >> A\test.txt
> svn commit -m test
> REM cherry pick test.txt change and commit to B
> svn up
> svn merge -r 2:3 A/test.txt B/test.txt
> svn commit -m test
> REM modify A/test.txt again
> echo >> A\test.txt
> svn commit -m test
> REM do an auto merge of B
> svn up
> svn merge file:///C:/test/test2/A <file:///C:\test\test2\A> B
> REM This produces merge infos in B only
> REM alternative
> svn revert B -R
> svn merge file:///C:/test/test2/A <file:///C:\test\test2\A> B --record-only
> REM This produces merge infos in B AND B/test.txt
> Regards,
> Stefan
>> Hi,
>> I'm wondering why there is a difference in how mergeinfos are
>> recorded based on whether the merge is done using --record-only or not.
>> To demonstrate the case, I've put together a repro-script (for
>> Windows - see attachment).
>> My question is that why does the last step in the script produce
>> different merge-info properties:
>> 1. svn merge file:///C:/test/test2/A <file:///C:\test\test2\A> B
>> This will produce mergeinfo in B
>> 2. svn merge file:///C:/test/test2/A <file:///C:\test\test2\A> B --record-only
>> This will produce mergeinfo in B and B/test.txt
>> Looking through the web, the docu and the SVN buglist I couldn't find
>> any matching entry. Maybe someone can point me on where to look for
>> an explanation?
>> I'm wondering especially because as an alternative to variation 2,
>> one might also follow variation 1 and then revert all changes (except
>> for the recorded mergeinfo B). Isn't the outcome then the same as
>> variation 2?
>> Regards,
>> Stefan

Received on 2015-02-24 18:27:51 CET

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.