Re: run pgindent on a regular basis / scripted manner
От | Jelte Fennema |
---|---|
Тема | Re: run pgindent on a regular basis / scripted manner |
Дата | |
Msg-id | CAGECzQTtsdX360pNP=nRoBpMcGwMK1aoYp4JgZwBMsMEKouKkQ@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: run pgindent on a regular basis / scripted manner (Andrew Dunstan <andrew@dunslane.net>) |
Список | pgsql-hackers |
> (ITYM "remove as many hurdles as possible"). yes, I messed up rewriting that sentence from "having as few hurdles as possible" to "removing as many hurdles as possible" > So far, we have had the following categories suggested: dirty, staged, dirty+staged, untracked. Are there any others? The two workflows that make most sense to me personally are: 1. staged (indent anything that you're staging for a commit) 2. dirty+staged+untracked (indent anything you've been working on that is not committed yet) The obvious way of having --dirty, --staged, and --untracked flags would require 3 flags for this second (to me seemingly) common operation. That seems quite unfortunate. So I would propose the following two flags for those purposes: 1. --staged/--cached (--cached is in line with git, but I personally think --staged is clearer, git has --staged-only but that seems long for no reason) 2. --uncommitted And maybe for completeness we could have the following flags, so you could target any combination of staged/untracked/dirty files: 3. --untracked (untracked files only) 4. --dirty (tracked files with changes that are not staged) But I don't know in what workflow people would actually use them. > Another issue is whether or not to restrict these to files under the current directory. I think we probably should, orat least provide a --relative option. Good point, I think it makes sense to restrict it to the current directory by default. You can always cd to the root of the repo if you want to format everything.
В списке pgsql-hackers по дате отправления: