Issue 131: Log viewer change proposal
Oleg Tazetdinov / TON on Fri, 27 Mar 2009 07:26:41 +0100:
It is not exactly a matter of large binary files. It is related to large text files as well.
I have a repository with about 1500 files. About half of them are sources and the rest are regression test results and some of them are about 4Mb in size. All are text files. It is generally very nice to use the log viewer to compare test results for different versions. The problem is the results get updated in a massive way, several hundreds at a time, and it takes very long time to generate the full diffs for such a changeset.
Besides the whole repository was imported into a new repository as a single changeset and it is rather dangerous to select this changeset in the log viewer:-)
I have two more comments to support my idea:
1. It takes the same amount (if not greater) of time to generate the diffs when committing such a large number of files but sometimes this is necessary (if strict control is required), in other times a command line mercurial can be used to do a commit. In any case commit is done once for each changeset.
2. The log viewer is quite different. It is used to analyse the changes and the same changeset can be revisited many times. I believe usually a user is interested in diffs for some specific files like when tracking some particular file history. Generating full diffs for a changeset in this case is just a waste of time and the diffs for a particular file coudl be generated on demand when a file is selected in the file box.
Probably a switch like a checkbox could be added somewhere to specify whether full diffs should be generated for each mouse click in the log tree.
It is just a proposal. It is up to you to decide whether my usage model is typical enough to justify all the trouble:-)
This is an issue notification from bitbucket.org.
You are receiving this either because you are the
owner of the issue, or you are following the issue.