Re: PKG_DEVELOPER=yes [Re: CVS commit: pkgsrc/lang/zig]

[Joerg Sonnenberger <joerg%bec.de@localhost>]

On Thu, May 14, 2020 at 07:42:20PM +0100, Jonathan Perkin wrote:
> * On 2020-05-14 at 18:30 BST, Nia Alarie wrote:
> 
> > Module Name:	pkgsrc
> > Committed By:	nia
> > Date:		Thu May 14 17:30:50 UTC 2020
> > 
> > Modified Files:
> > 	pkgsrc/lang/zig: Makefile
> > 
> > Log Message:
> > zig: Installs a script that requires bash.
> > 
> > Replace the interpreter so this doesn't fail with PKG_DEVELOPER.
> > 
> > Please set PKG_DEVELOPER=yes to detect problems like this!
> 
> We should just set this by default.  There's absolutely no excuses for
> packages to be failing these basic checks.
> 
> Maybe we move all of the current tests outside of the PKG_DEVELOPER
> checks, and instead make =yes to mean much more stringent additional
> checks, like some of the stuff Roland is working on, ensure pkglint
> runs clean, and turning on some additional CHECK_WRKREF etc.

I'd be a bit more conservative. CHECK_INTERPRETER can always be enabled.
CHECK_PERMS pulls in an extra dependency and I wouldn't put that much
value in it. Not sure about CHECK_FAKEHOME. CHECK_SHLIBS also seems
useful enough to have by default. All that said, I really hope that
PKG_DEVELOPER is enabled by everyone before commit, no excuses. It
should also be used for all bulk builds.

Joerg