bug#59921: 30.0.50; emacs HEAD is missing some older emacs-29 commits

[Brian Leung]

Eli Zaretskii <eliz@gnu.org> writes:

> > Cc: 59921@debbugs.gnu.org
> > From: Brian Leung <leungbk@posteo.net>
> > Date: Fri, 09 Dec 2022 12:43:12 +0000
> >
> >
> > > Doesn't this mean that these commits are on the master 
> > > branch?
> >
> > I arrived at a similar conclusion using Magit. I don't have a 
> > great understanding of Git, so my best guess is that 
> > 64c7b57ae72 
> > somehow skipped over more commits than expected. You can verify 
> > that the changes corresponding to the missing commits are 
> > present 
> > in the emacs-29 branch but not in master.
>
> How did you detect the missing changes in the first place?

To be extra clear about what I think happened, 
64c7b57ae726f6483921db5f6daeda8ada162666, which was a merge from 
emacs-29, has a commit message that reads:

> The following commit was skipped:
>
> 508b0c1b22 Fix crash on Windows 9X

The tree-sitter-related changes I mentioned earlier in the thread 
are not present in 64c7b57ae726f6483921db5f6daeda8ada162666.

However, if you check out 508b0c1b22, you will find that the 
tree-sitter-related changes indeed exist there. In the emacs-29 
branch, 508b0c1b22 is a descendant of the tree-sitter-related 
changes, so any merge commit that merged 508b0c1b22 into master 
ought to have merged any ancestors of 508b0c1b22 as well. But 
508b0c1b22 was intentionally skipped, and some of its ancestors 
seem to have been skipped by mistake as well.

508b0c1b223 | * | Fix crash on Windows 9X
670daa8b626 | * | Add recursion limit to 
treesit--children-covering-range-recurse
b429e524288 | * | Improve tree-sitter fontification in edge-cases 
(bug#59738)
c83c95634e7 | * | Add c-ts-mode-indent-defun (bug#59662)
6479691cf07 | * | Fix syntax-table for tree-sitter modes
b710ca62c00 | * | c++-ts-mode: Highlight nullptr as a constant
d31a2539834 | * | Improve parameter checking in tree-sitter 
functions
a669d5fae54 | * | ; * test/lisp/comint-tests.el 
(w32-application-type): Fix typo.
f7262b8f81e | * | Fix comint-tests on MS-Windows
fef17557365 | * | Use new external-completion.el in Eglot
bf81df86e52 | * | Don't override completion-category-defaults in 
fido-mode
4bc9c8972ac | * | Add lisp/external-completion.el for completions 
from external tools
bfc00f1c120 | * | Eglot: fix setting of 
flymake-list-only-diagnostics (bug#59824)
d2411615e8b | * | * test/src/sqlite-tests.el (sqlite-version): 
Declare.

A few of the above commits had their corresponding changes make 
their way into the current master (for example, 4bc9c8972ac, which 
added the external-completion stuff). However, fef17557365, which 
began to use the external-completion.el package in eglot, is 
present in emacs-29 but not in master. I cannot see any mention of 
external-completion package when looking at eglot.el in the master 
branch, whereas I see it prominently mentioned in the 
Package-Requires when I check out the emacs-29 branch.