Jan Vrany <jan.vrany@labware.com> [Mon, 26 Oct 2020 11:01:31 +0000] rev 915
Correct method comment in `HGMergeToolStartup >> #main:`
Jan Vrany <jan.vrany@labware.com> [Thu, 08 Oct 2020 12:52:31 +0100] rev 914
Rename "temporaryWorkingCopy" to just "workingCopy" thorough the code
This is a preparation for removing use of temporary working copy
altogether. The main problem is that it causes confusion, mainly
among newcomers.
Jan Vrany <jan.vrany@labware.com> [Thu, 08 Oct 2020 18:33:52 +0100] rev 913
Use separate logic to detect classes whose copyright should be updated
While `#shouldFileOutClass:` is conservative, i.e., may return true for classes
that were not changed, (new) `#shouldUpdateCopyrightInClass:` should never return
`true` in that case, the opposite (i.e., returning false in cases when class has
in fact been changed) is unfortunate but somewhat acceptable.
Jan Vrany <jan.vrany@labware.com> [Thu, 08 Oct 2020 18:03:13 +0100] rev 912
Fix `HGChangeset >> #/' to work with obsolete changesets
Jan Vrany <jan.vrany@labware.com> [Thu, 08 Oct 2020 17:58:13 +0100] rev 911
Update .hgignore
Jan Vrany <jan.vrany@labware.com> [Thu, 27 Aug 2020 17:37:29 +0100] rev 910
Introduce `HGMergeTool` - a merge tool for Smalltalk/X source files
`HGMergeTool` merge tool support another (Mercurial specific)
invocation:
... -r REV LOCAL
This can be used to cherry-pick changes to `LOCAL` file from revision
`REV`. It automatically searches history for base revision to guide
the (automatic) merge.
Jan Vrany <jan.vrany@labware.com> [Thu, 27 Aug 2020 09:29:13 +0100] rev 909
Introduce `SCMMergeTool` - a merge tool for Smalltalk/X source files
This can be used by SCM (such as Mercurial or GIT) to merge Smalltalk/X
source files (.st) instead of standard 3-way merge tools (such as `kdiff3`
or `meld`). The advantage is that it operates atop of changesets, not
on whole files.
Jan Vrany <jan.vrany@labware.com> [Wed, 19 Aug 2020 13:03:08 +0100] rev 908
Fix `HGTests >> #test_hgCommand_03_unix` for Python3
Use the same Python as the Mercurial's `hg` by reading `hg`'s shebang.
This is to handle cases where Mercurial is using Python 3 but `python`
command refers to (old) Python 2.
Jan Vrany <jan.vrany@labware.com> [Mon, 29 Jun 2020 23:53:54 +0100] rev 907
Make method's source local (i.e., in-memory) when filing-out class
...such as when commiting. This is a measure against source corruption
when filing out to package directory (this is not yet supported,
but this commit in a preparation for that)
Jan Vrany <jan.vrany@labware.com> [Tue, 30 Jun 2020 00:33:00 +0100] rev 906
Fix `HGStXTests >> #test_commit_nested_07a`
The aim of this test is to check that changes are correctly
shrunk only for commited classes / extensions.
However, it incorrectly checked whether current changeset does not
contain changes for class NOT commited. This is wrong - package commit
may have touched other classes (like, adding copyright) but unless
really commited, changes should not be shrunk for these uncommited
classes!
Interesting how long this flaw in this test when unnoticed. This has
been uncovered by fixing / improving copyright update code.