Re: [PATCH v3 13/28] t3702: abstract away SHA-1-specific constants

From: brian m. carlson <sandals@crustytoothpaste.net>
Date: May 16, 2018, 23:46 UTC
Message-ID: <20180516234648.GC652292@genre.crustytoothpaste.net>
In-Reply-To: <CAGZ79kYrDDGtSSViUwqOaJCTUUcUaiJe1DQHGSqS+3SCYnEv8g@mail.gmail.com>

On Wed, May 16, 2018 at 10:18:33AM -0700, Stefan Beller wrote:

Show 10 quoted lines

> This reminds me of the way we test alot of the patch format already.
> But there we use standard grep as opposed to egrep.
> 
> git grep egrep doesn't show a lot of hits, but all commits
> that mention egrep found via 'git log --grep egrep' mention
> that there is some sort of portability issue for using egrep
> specifically.
> 
> Is the ^index a problem for standard grep, i.e. do we need to fix
> other places?

I think this is just me preferring a more careful match, but if there are potential portability problems, I can reroll to use regular grep. I don't think which one we use matters much one way or the other in this case, or in general. We don't tend to produce a lot of potentially ambiguous matches in our testsuite.

I think the uses in the commit messages tend to point out quirks in command line options over specifically concerns about the use of egrep itself. I suspect the implementations that want egrep over grep -E (the latter being POSIX) also lack many of the POSIX options that people want to use, although I'm not aware of egrep itself being broken anywhere.

-- 
brian m. carlson: Houston, Texas, US
OpenPGP: https://keybase.io/bk2204