qlog: Expand to found revisions when searching
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
QBzr |
Confirmed
|
Medium
|
Unassigned |
Bug Description
Hello, I'm using the latest qbzr pulled today, and a recent bzr.dev.
I have this branch: https:/
with <email address hidden> as latest revision.
I do "bzr qlog", and in the search field I enter "1110" (seaching in Messages).
It displays a list of revisions: 2755 2730 2720 2716 2710 2498 1873.
None of those contains "1110" in its message. Actually they are all merge revisions, and they merged a revision containing "1110" in its message. Seeing only the merge revisions displayed is a problem, because the only solution (to find the revisions actually having "1110" in their message) is to expand each merge revision (clicking on the "+"), for example expanding 2498: this expands to 6 other merge revisions, which I also have to expand, etc. After expanding everything (which requires 52 clicks on "+" signs), I finally see the list of all (non-merge) revisions containing "1110" in their message, which is what I was after.
Contrast this with when searching for revids, for example for
<email address hidden>
which automatically expands merge revisions as needed and highlights the found revision (no expanding needed).
I understand one reason for the search-by-message behaviour: whereas searching by revid is selective, searching by message may not be very selective, and thus auto-expanding could produce a huge graph stalling qlog.
Are there solutions to this? Keep things unexpanded but offer an "expand all" button?
Same effect is observed when searching by author.
tags: | added: mysql |
summary: |
- "bzr qlog" displays revision when searching in revids, but not when - searching in messages + qlog: Expand to found revisions when searching |
Changed in qbzr: | |
status: | Incomplete → Confirmed |
importance: | Undecided → Medium |
tags: | added: qlog |
Workaround for now:
Try running "bzr qlog --no-graph" this will show all revisions in expanded state and stacked one on top of the other.
Not sure if this workaround is good enough?