[cfe-dev] [LLVMdev] RFC: put commit messages in *-commits subject lines?

Sean Silva silvas at purdue.edu
Thu Nov 15 23:28:32 PST 2012


> I agree that the list name is redundant and should be dropped, but the revision number is compact and very useful...

It's the only part of the message that says "r123456", including the
'r', which in my experience makes it *the* go-to for copypasting.

-- Sean Silva

On Thu, Nov 15, 2012 at 10:40 PM, Chris Lattner <clattner at apple.com> wrote:
> I agree that the list name is redundant and should be dropped, but the revision number is compact and very useful...
>
> -Chris
>
> On Nov 15, 2012, at 6:45 PM, Tobias Grosser <tobias at grosser.es> wrote:
>
>> On 11/16/2012 02:47 AM, Chris Lattner wrote:
>>>
>>> On Nov 15, 2012, at 2:28 PM, Chandler Carruth <chandlerc at google.com> wrote:
>>>
>>>>> [cfe-commits] r167788 - in /cfe/trunk: lib/Analysis/CFG.cpp test/Analysis/dead-stores.cpp
>>>>>
>>>>> with
>>>>>
>>>>> [cfe-commits] r167788 - Fix bad CFG construction bug when handling C++ 'try' statements.
>>>>>
>>>>> The first gives me just enough information (in my abbreviated mesage list window) to see that there was a change to Clang's Analysis library; if I look at the whole subject, I can see what files were changed (or the first few, at least). But the second actually tells me what changed, and whether or not it affects me.
>>>>>
>>>>> Having this form will also encourage people to put a useful summary in the first line of their commit, which is useful for anyone viewing history (especially git users, where the one-line summary is very common).
>>>>>
>>>>> Is there a reason we currently prefer files to log messages?
>>>>
>>>> I suspect not. I would much prefer your suggestion. If you can
>>>> implement the fix to our scripts (and get Tanya or Chris or Anton to
>>>> give you access), I would be thrilled.
>>>
>>> I'm in favor of it.  Of course, the truly awesomest thing would be something like:
>>>
>>>     [cfe-commits] r167788 - [lib/Analysis] Fix bad CFG construction bug when handling C++ 'try' statements.
>>
>> Moving the important information first may also be worth considering. Subject lines are often cut off.
>>
>> I prefer
>>
>> [lib/Analysis] Fix bad CFG construction bug when handing C++ 'try' ...
>>
>> rather than
>>
>> [cfe-commits] r167788 - [lib/Analysis] Fix bad CFG construction ...
>>
>> "[cfe-commits]" can be removed from the subject as the amount of mails forces people anyway to filter/tag their emails
>>
>> We can leave "r167788" in the subject line as people probably search for it. However, the commit message seems to be better to actually recognize a certain commit. So we do not need the revision at the beginning of the line.
>>
>> Cheers
>> Tobi
>>
>>
> _______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev




More information about the cfe-dev mailing list