Hello Richard, I implemented this feature years ago, but I only had it in my branch 'annotate_links'. It was refused among with other changes that at some point I had added to trunk. Much later, you implemented the diff clicking graph nodes.
The point of the "branch diff" is to see what would this branch bring to the parent branch, if merged (integrated). This works for our workflow, which means that branches created from parent A are only "updated" from parent A. If we create branch "fix" from trunk, then we only do merges from trunk, and we will integrate it only to trunk, if ever. If we create branch B from A, then we only update B from A, and we end up (if ever) merging B to A. Often, our timeline is long or complex enough so the click of two nodes is quite annoying. Moreover, the "branch diff" does for us the right thing without even having to find checkins, where we could choose wrong. I don't understand the workflow of http://core.tcl.tk/tcl/timeline?r=bug-1224888 . How will this bug-1224888 branch be integrated? Integrated to what? As for http://www.sqlite.org/src/timeline?p=cdaeab4, it diffs from [42d5601739]. That shows what would ota-update would bring to trunk if merged. So, it looks right to me. One thing I was lazy to write was to disable the button in case there is no "parent branch", like what would happen with trunk checkins. Regards, Lluís. > ---------- Forwarded message ---------- > Lluis and Joe, I want to give y'all an opportunity to "sell" me on the > "branch diff" link that you added to the /ci and /vinfo pages and > implemented on the "viric_pbranch" branch of Fossil: > > https://www.fossil-scm.org/fossil/timeline?r=viric_pbranch > > My concern is that when you do a "branch diff" it is not clear what > you are diffing against. Take for example, the bug-1224888 branch > over on the Tcl project: > > http://core.tcl.tk/tcl/timeline?r=bug-1224888 > > Experimentally we see that clicking "branch diff" button on the tip of > bug-1224888 does a diff against check-in 690300420f from trunk, not > check-in 80c27dee1f from core-8-4-branch. Is that was is intended? > What about something like check-in cdaeab4 in SQLite > > http://www.sqlite.org/src/timeline?p=cdaeab4 > > In other words, what is the algorithm that you use to determine the > start of a branch? > > As you can see, the 'start of a branch' can sometimes be a confusing > concept. Wouldn't it be better and easier for the user to use the > click-on-two-graph-nodes technique to select the precise two check-ins > he wants to diff, rather than relying on the "branch diff" link to try > to figure out a reasonable base version based on heuristics? > > -- > D. Richard Hipp > d...@sqlite.org > _______________________________________________ > fossil-dev mailing list > fossil-dev@lists.fossil-scm.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/fossil-dev -- PGP key D4831A8A - https://emailselfdefense.fsf.org/ _______________________________________________ fossil-dev mailing list fossil-dev@lists.fossil-scm.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/fossil-dev
