♟ martinvonz

rename: add --forget option and stop suggesting `hg revert` for undoing

tests: avoid using rebaseskipobsolete=0

rebase: when using --keep, don't care about pruned commits or divergence

rebase: remove duplicate initialization of a field

Seems uncontroversial, so I'll take it despite the fact that the patch is from a coworker.

tests: update divergence test for `hg fix` to actually result in divergence

rebase: don't call rewriteutil.precheck() with to-be-skipped commits

reabase: call rewriteutil.precheck() a bit later

rebase: set `prepared = True` at very end of `_preparenewrebase()`

rebase: filter out descendants of divergence-causing commits earlier

rebase: clarify names of variables and function related to obsolete revisions

rebase: calculate obsolescense-related info earlier

rebase: let _handleskippingobsolete(self) read directly from self

tests: test divergence created during interrupted rebase

tests: split up test-rebase-obsolete.t in four pieces

In D10249#156144, @martinvonz wrote:

I may make some changes to this patch. I keep getting confused about the two different divergence cases. Despite have stared at this code for days, I still don't know which case is which (i.e. I'm not sure my description of cases (3) and (4) is correct).

I may make some changes to this patch. I keep getting confused about the two different divergence cases. Despite have stared at this code for days, I still don't know which case is which (i.e. I'm not sure my description of cases (3) and (4) is correct).

fuzz: use Python 3 in makefile

makefile: use Python 3 by default when building docs as well

tests: replace some `python` by `"$PYTHON"` to not only use Python 2

rebase: skip obsolete commits even if they have pruned successors

bisect: use standard one-line commit summary

This made test-git-interop.t fail more than before. Can you send a fix? The end of the traceback looks like this:

amend: mark commit obsolete after moving working copy

split: close transaction in the unlikely event of a conflict while rebasing

rebase: inline simple function for finding obsolete subset of commits

In D10115#154171, @marmoute wrote:

Yeah I meant 'a', sorry about that (the 'd' comes from a copies discussion I had with Pulkit in //). Line 192 does not help me much since we still seems to be sorting the longest first as far as I undersant (which result in the deepest coming first).

In D10115#154127, @martinvonz wrote:

In D10115#154126, @marmoute wrote:

Is this also going to affect case where entry are nto childre of each other but still have different side ? for example if we had ['a/short', 'a/longer'] with a prefix matching on 'd' this change alter the behavior, don't it ?

I don't understand. d doesn't match either of those. Did you mean a? Even assuming that, I'm not sure what you mean, but maybe line 1092 answers your question?

In D10115#154126, @marmoute wrote:

Is this also going to affect case where entry are nto childre of each other but still have different side ? for example if we had ['a/short', 'a/longer'] with a prefix matching on 'd' this change alter the behavior, don't it ?

copies: filter out copies grafted from another branch

copies: inline _backwardrenames() in pathcopies()

copies: extract function _backwardcopies() for reversing renames

tests: demonstrate how grafted copies are counted when tracing across branches

This patch makes test-releasenotes-formatting.t fail for me. I noticed that it has #require fuzzywuzzy at the top. Could you install that (I don't even remember what it is) and send a fix?

copies: choose target directory based on longest match

martinvonz (Martin von Zweigbergk)
User

Projects

User Details

Recent Activity
View All

Sun, Apr 11

Fri, Apr 9

Thu, Apr 8

Tue, Mar 30

Mon, Mar 29

Fri, Mar 26

Thu, Mar 25

Wed, Mar 24

Tue, Mar 23

Mon, Mar 22

Sat, Mar 20

Fri, Mar 19

Tue, Mar 16

Mar 13 2021

Mar 12 2021

Mar 9 2021

Mar 8 2021

Mar 5 2021

Mar 4 2021

martinvonz (Martin von Zweigbergk)User