Item9969: REVINFO and $meta->getRevisionInfo() has issues when the ,v file is re-built after it was lost.

this shows up in the REVINFO macro in 1.1, but only affects getRevisioninfo in 1.0, so is a regression in 1.1

if you have a topic that has a rev > 0, and delete its ,v file, and then edit and save again, you get a new ,v file.

however, the revision history is a bit messed up. The ,v file will start fresh r1 with the .txt file, and then the save will be r2

# lets say that the metaINFO when we lost the ,v file was rev=4
my ($meta, $text) = Foswiki::Func::readTopic($web, $topic, 2);
my ($date, $author,$rev,$c) = meta->getRevisionInfo()= $meta->getRevisionInfo()
#rev == 2
my ($meta, $text) = Foswiki::Func::readTopic($web, $topic, 1);
my ($date, $author,$rev,$c) = meta->getRevisionInfo()= $meta->getRevisionInfo()
#rev = 4
my ($meta, $text) = Foswiki::Func::readTopic($web, $topic, 4);
my ($date, $author,$rev,$c) = meta->getRevisionInfo()= $meta->getRevisionInfo()
#rev = 2

imo that the rev returned by getRevisionInfo does not match is annoying, but slightly reasonable (it'd be better if the missing ,v file resulted in the first rev == 4, and the missing revs where undefined / empty..

but that readTopic of a future revision silently succeeds i think is bad.

-- SvenDowideit - 07 Nov 2010

If the ,v file is removed the revisions should start over from 1.

This is essential and I use this all the time when I make new applications by copying some old web at file system level. Then I copy and deleted the ,v files. It would be a major pain in the arse if I also have to hack the revision

And the number of support calls I will geg when a topic is rev 4. "where are revs 1-3?" will be the question over and over again.

No let the revision start at 1 when the ,v file is deleted.

And it is not a release blocker for 1.1.2 no matter what.

-- KennethLavrsen - 07 Nov 2010

This has been changed a bit Item10307. If rcs file is missing, the initial rev is rev 0. First save will create rev 1 for the existing topic in store, AND rev 2 for the initial changes. Bumped target release to 1.1.4 - no fixes committed to 1.1.2.

-- GeorgeClark - 12 Mar 2011

Kenneth is right, the history should start again from 1. There is no way to distinguish between a lost history, and a history that never existed e.g. a manual edit that happened to include TOPICINFO. The only rational course is to start again from 1.

Returning a rev 0 is IMHO a mistake, though I'm not going to start breaking 1.1.4 to assert that.

See Item10091 for what I'm doing on trunk.

-- CrawfordCurrie - 31 Aug 2011

I think this is no action - Revisions start over from Rev 1 now. That is the desired behavior

-- GeorgeClark - 13 Dec 2011

Commenting is disabled while not logged in