Re: [PATCH 1/3] docs/zh_CN: add process/applying-patches Chinese translation

[Weijie Yuan]

Sorry I mistakenly dropped alexs@xxxxxxxxxx, add it back.

(Is the forwarding address of kernel.org gmail.com? I was worried about
sending duplicate emails. So I removed kernel.org)

Anyway, sorry for the noise.

On Sun, Jul 26, 2026 at 03:11:23PM +0800, Weijie Yuan wrote:
> +cc Chen-Yu
> 
> Hi Alex,
> 
> On Sat, Jul 25, 2026 at 07:59:37AM +0800, Alex Shi wrote:
> > On 2026/7/18 00:43, Weijie Yuan wrote:
> > > On Fri, Jul 17, 2026 at 08:07:42PM +0800, Weijie Yuan wrote:
> > > > Translate .../process/applying-patches.rst into Chinese.
> > > > 
> > > > Update the translation through commit fa04150b8ef7
> > 
> > Is it the original meaning: translation updated to commit xxxx?
> 
> My bad. I might misunderstand "update through" and "update to".
> 
> I originally thought that "update through" could only mean that this
> current Chinese commit only updated the specific content from that
> previous English commit.
> 
> Thanks to Chen-Yu, he taught me a lot:
> 
> In tools/docs/checktransupdate.py, the function
> get_origin_from_trans_smartly() uses two regular expressions to extract
> the reference point:
> 
>     update to commit <12bits-hash>
>     Update the translation through commit <12bits-hash>
> 
> So both of them works.
> 
> But later he mentioned a few potential gotchas which perhaps would be
> worth leaving a note here.
> 
> 1. The regex does not enable IGNORECASE, so "update to commit" must be
>    written in lowercase. If it is written as "Update to commit" with an
>    uppercase U, the script will not recognize it and will fall back to
>    using the author date to guess the base commit.
> 
>    This is particularly subtle because the script does not report an
>    error; it simply produces an incorrect result silently.
> 
> 2. The script only checks how many commits the translation is behind
>    relative to the hash you specified. It does not verify whether that
>    hash is actually the latest commit touching the corresponding English
>    file.
> 
>    As a result, if you accidentally specify an older hash, the script
>    may instead report that the translation is "up to date." Before
>    sending his latest series, Chen-Yu manually checked every file with:
>    git log -1 -- <English file>
> 
> 3. valid_commit() skips merge commits, so a merge commit cannot be used
>    as the base. 
> 
> Sorry I have not yet dug into the script in detail. So if any of the
> points above are already documented there, I apologize for having
> overlooked them. If they are not, we could add them to howto.rst, or
> take a look at how the script might be improved when time permits.
> 
> Thanks!