D1551: remotenames: consider existing data while storing newer data
durin42 (Augie Fackler)
phabricator at mercurial-scm.org
Wed Dec 6 13:55:49 EST 2017
durin42 added a comment.
In https://phab.mercurial-scm.org/D1551#27331, @grim wrote:
> As an outsider looking in, I can see both points of view here. But as someone maintaining a large code base here I have to say I agree with Sean. The whole "we'll fix it later thing" is crap. It never happens as long as the code "works good enough". Case in point, I'm currently systematically removing 228 #if 0's from the Pidgin code base, most of which were put there more than a decade ago.
That is, frankly, the point: smf has been requesting reasonable improvements in a thing that is already working and widely deployed that are then resulting in us shipping *nothing*.
> In my experience there are two times when code in this situation can/will be cleaned up. Before it's merged or when someone working on a bug fix/feature get's completely frustrated with it's state and ends up refactoring the crap out of it. Right now this feature is at the former but this discussion seems to be leading to the later, which is actively buying into the long term maintenance cost. Perhaps Mercurial can afford that cost, but this is not something I would advise/condone on my own projects.
My expectation is that I have two realistic choices: reject the feature I want, or take it with a hope that someday someone improves it to be better than it is today. I don't believe there's a path (as you seem to?) that involves someone meeting a higher bar to get this feature shared to all hg users in the next year.
To: pulkit, #hg-reviewers, durin42
Cc: grim, smf, durin42, dlax, mercurial-devel
More information about the Mercurial-devel