[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Is it time to drop ChangeLogs?

From: Karl Fogel
Subject: Re: Is it time to drop ChangeLogs?
Date: Wed, 09 Mar 2016 14:42:30 -0600
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/25.0.92 (gnu/linux)

Eli Zaretskii <address@hidden> writes:
>I don't even need to explain that: there are already people in this
>discussion who said they'd like to use a less formal format.  (You
>already asked who those people were, and Dmitry already pointed out
>who they were.)  To me, the connection is very clear.  If you don't
>see it, just trust the facts -- it is already happening.

I've worked in projects where we successfully enforce ChangeLog-style commit 
message conventions without having actual ChangeLog files.  If the Emacs 
project is having a social problem enforcing conventions, that's certainly 
something to be solved.  But blaming it on the absence of ChangeLog files does 
not hold up empirically.

I do not see any connection between the fact that some people advocate using a 
less formal format, and the question of whether or not we should have ChangeLog 
*files*.  One is about commit message format (other projects enforce a format 
successfully, and some even enforce this particular format), the other is about 
whether certain files exist in the tree.   It's surprising to me that you think 
this connection is so obvious that you don't even need to explain it :-), 
whereas to me there is no strong connection at all.

>> You wrote of the importance of "ChangeLog files" to forensics, but
>> having that information in git commit messages is exactly equivalent
>> (I use it for forensics all the time, in lots of projects).
>Once again, you take a part of the argument and try to analyze it
>The point was that if we start on the slippery slope of saying the
>mistakes in the log messages are unimportant, we will eventually lose
>both the information in the Git log records and the ability to produce
>ChangeLog files.
>IOW, it's fortress -- if you dismantle one of its bastions, the others
>will soon fall as well.  The only way to avoid that is to hold all of

Okay, I understand your argument (I don't think your previous messages made it 
clear, but you've clarified it now).  However, it just does not hold true in 
other projects.  Why would we expect Emacs developers to be more slovenly than 
developers elsewhere?

>> Do you see now why it at least looks like you're conflating these two 
>> different things?
>No.  There are different aspects to this issue, and I described them
>one after the other.  Sometimes they are only loosely related,
>sometimes they are more tightly coupled.  This is the best way I could
>express my thoughts.  Since English is not the first language for
>either of us, let's just agree that I failed to write in a style that
>would allow you to better understand.

English is my native language, by the way; I'm American, not German.  FWIW, I 
couldn't tell from your writing that English is not your native language.

>Once again, if there's a way to have the information in Git such that
>a correct log could be generated from it, it would be fine with me,
>and keeping the ChangeLog files in the repository won't be necessary,
>as far as I'm concerned.  But so far, keeping the files looks like the
>cheapest way of satisfying all the requirements.  I understand the
>psychological effect of "going the old ways", but this is a practical
>matter for me, not a religious one, and I'm looking for a practical
>solution that would allow us to continue keeping an accurate accord of
>the development history with as little overhead as possible.  If the
>best alternative is those "old ways", I have no problem with that.

I think this project can successfully enforce ChangeLog-style entries in commit 
messages if we really want to, without much difficulty.  There are a number of 
techniques for doing so, mostly social techniques, with a few technical helpers 
if we want.  Other projects with similarly cohesive developer communities 
enforce such conventions all the time.  I don't think it has anything to do 
with whether or not there are actual ChangeLog files in the repository; the 
fact that some of the people who don't want to keep those files are *also* 
people who don't want to use that style of entry generally, even in commit 
messages, is merely correlative -- getting rid of the files doesn't mean that 
everything else they advocate must happen too.

I believe this is also what Paul was saying, though I don't want to put words 
in his mouth.

Best regards,

reply via email to

[Prev in Thread] Current Thread [Next in Thread]