On 2016-04-26 12:09 PM, Ævar Arnfjörð Bjarmason wrote:
Basically you can look at patches to a project on a spectrum of: 1. You hacked something up locally 2. It's in someone's *.git repo as a POC 3. It's a third-party patch series used by a bunch of people 4. In an official release but documented as experimental 5. In an official release as a first-rate feature Something like an experimental.WHATEVER=bool flag provides #4.
But the git project already does #4 without needing a special configuration tree. In fact, you ignored the git project's "pu" branch entirely. Once a feature gets onto the "next" branch, it's already much less experimental. If it needs to put the word "experimental" in its config settings, then maybe shouldn't leave the "pu" branch in the first place.
I think aside from this feature just leaving these things undocumented really provides the worst of both worlds.
Yes, I apologize. I did not mean that things should remain undocumented, only that if you're afraid of users harming themselves then you're better off not documenting settings than labelling them as experimental.
Now you have some feature that's undocumented *because* you're not sure about it, but you can't ever be sure about it unless people actually use it, and if it's not documented at all effectively it's as visible as some third-party patch series. I.e. only people really involved in the project will ever use it.
Slapping "experimental" on the configuration only serves to muddy the waters. Either the feature is good enough to be tried by normal users, or it isn't. If it isn't ready for normal users, let it cook in pu (or next) for a while longer.
Git has got on just fine without having some special designation for not-ready-for-prime-time features, mostly because the development process lends itself naturally to gradually exposing things as they mature: topics move from the list to pu to next to master. (The string "experiment" only appears 16 times in all the release notes, which I think is something the git project can be proud of.)
To me, designating part of the config tree as "experimental" enables sloppier practices, where a feature can be released with a bit less effort than might otherwise be acceptable, because it's clearly marked as experimental, and so anyone trying it out surely has the requisite bulletproof footwear. (I don't mean to imply that you or any other git contributor might slack off on any work you do for the project. It's more that the ability to easily designate something as experimental lowers the bar a bit, and makes it more tempting to release not-quite-ready features.)
Far better to instead work on the feature until it's as ready as can be, and release it properly.
In this particular case, for example, I think your proposed approach is perfectly fine and does not need to be designated as "experimental" in any way. With a reasonable "hooks.something" config variable to turn on directory support, what you've described sounds like a great feature. Sure, it may have bugs, it may have unintended consequences, it may not fulfill some odd corner cases. But that's true for almost everything. As with every other git feature, it'll be developed to the best of the project's abilities and released. Future patches are welcome.
M. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html