Re: [RFC PATCH v3 1/1] unit tests: Add a project plan document

From: Linus Arver <linusa@google.com>
Date: Jun 30, 2023, 00:11 UTC
Message-ID: <owly7crl95c3.fsf@fine.c.googlers.com>
In-Reply-To: <xmqqedlu7yn2.fsf@gitster.g>

Junio C Hamano <gitster@pobox.com> writes:

Show 12 quoted lines

>> I can think of some other metrics to add to the comparison, namely:
>>
>>     1. Age (how old is the framework)
>>     2. Size in KLOC (thousands of lines of code)
>>     3. Adoption rate (which notable C projects already use this framework?)
>>     4. Project health (how active are its developers?)
>>
>> I think for 3 and 4, we could probably mine some data out of GitHub
>> itself.
>
> Great additions (if we are mere users do we care much about #2,
> though?).

Sorry, I forgot to add why I think these metrics are useful: I think they give some signal about how much influence/respect the framework has in our industry, with the assumption that the influence/respect positively correlates with how "good" (sound architecture, well-written, easy to use, simple to understand, etc) the framework is. For the frameworks hosted in GitHub, perhaps the number of GitHub Stars is a better estimate for measuring influence/respect.

That said, I think #2 (measuring KLOC) would still be useful to know (and is easy enough with tools like tokei [1]), mainly for the scenario where the framework becomes abandonware. Certainly, a framework with a lower KLOC count would have a lower maintenance burden if we ever need to step in to help maintain the framework ourselves. To me this is one reason why I like the idea of using Phillip Wood's framework [2] (granted, it is currently only a proof of concept).

[1] https://github.com/XAMPPRocky/tokei [2] https://lore.kernel.org/git/c902a166-98ce-afba-93f2-ea6027557176@gmail.com/