Re: [lmi] [lmi-commits] master 0d8f53e 02/18: Resolve shellcheck "SC1117" warning

[Vadim Zeitlin]

On Thu, 30 May 2019 13:19:26 +0000 Greg Chicares <address@hidden> wrote:

GC> In exchange for restricting myself to the subset of 'sh' code that
GC> shellcheck accepts without warning, I get notified automatically
GC> of some mistakes, before I even commit them. For me, that's a good
GC> tradeoff.

 If it did find some real errors for you, then it's indeed good. But my
problem with shellcheck is that so far I haven't seen it detect any real
problems (some of its warnings are fair enough and _could_ result in some
trouble if the scripts were used with unconstrained inputs, e.g. if we used
paths with spaces in them -- but we don't, and hopefully never will) while
it costs a rather inordinate amount of time to fix or work around them (I
especially liked when its own advice for working around one of its warnings
resulted in another warning).

 I love linters myself, but I think they need to have either very few false
positives or detect sufficiently many genuine problems to be more helpful
than aggravating and so far shellcheck doesn't fall on the right side of
this divide for me.

</end-of-rant>

VZ

pgpRM7VYaIxnR.pgp
Description: PGP signature