pkgsrc-Changes archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: CVS commit: pkgsrc/lang/racket-textual

On 9/24/2012 20:20, Aleksej Saushev wrote:
John Marino<>  writes:

On 9/24/2012 13:18, Aleksej Saushev wrote:
Module Name:    pkgsrc
Committed By:   asau
Date:           Mon Sep 24 11:18:38 UTC 2012

Modified Files:
        pkgsrc/lang/racket-textual: Makefile

Log Message:
Revert changes that make no sense.

To generate a diff of this commit:
cvs rdiff -u -r1.11 -r1.12 pkgsrc/lang/racket-textual/Makefile

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.

Translation: Alexsej just changed all the "{}" back to "()" for
no benefit, and to apparently to intentionally cause pkglint to
generate warnings for the reason I can only speculate that he
thinks pkglint should not classify "()" as an error.

This revert had no justification and it makes the pkgsrc
Makefile worse (same story with lang/racket).   This is
irresponsible.  For a lesser extent, OBATA Akio pointed me to
commit guidelines that said one should avoid reverting commits
of other developers and that was violated as well.

Apparently there's no point on using pkglint, at least on
packages that asau has interest in.

This reversion has benefit to me as maintainer at the very least.

pkglint is broken, and I have raised this problem previously.
First and the biggest problem with pkglint is that it insist on
undocumented rules with unclear reasoning (sometimes harmful ones)
which brings it into category of "probably good, if used with care"
at most rather than a tool to remove mistakes.
Thus instead of following all pkglint suggestions blindly, please,
think about the reasoning behind all what happens there.

You have not consulted with me about changing these packages, even though
I maintain them and I use them. In addition, there're more packages that
follow these. All that you have made is introduction of pointless and
harmful (at least, potentially) changes that cause problems to me as
package maintainer.

The same applies to using non-tested JPEG library (I have never tested
Racket with that another option and have no time to test it for now)
and using LOCALBASE (which is dubious change as well, though it might
have some sense which is why I didn't revert it).

It is not the first time you introduce changes that have very little
reason behind them (if any at all), sometimes they have only aesthetic
value and sometimes are just harmful ad-hockery. In future, please,
provide clear reasoning. Otherwise, I'll have to revert non-functional
changes that get in a way rather than serve any purpose in packages
I maintain.

I know for a fact that using "()" instead of "{}" causes some pkglint logic to fail. All things being equal, "{}" is therefore superior to "()" where pkglint is concerned. Rather than have to reprogram pkglint for both "()" and "{}" cases, it makes perfect sense to flag one version as "bad". This doesn't even address consistency which is another valid reason to allow only one case.

It's one thing to not bother to change them on principal. It's quite another to go out of your way to revert a legitimate commit in order to make the Makefile "wrong" in the eyes of pkglint.

This commit is nothing more than childish protest against pkglint. pkglint is an official pkgsrc tool, and cleaning up pkglint warnings and messages is a valid reason in an of itself. Obviously if you claim most of these changes are aesthetic, you are clearly admitting they make no difference either way. And if you have such a mad-on about pkglint, fix it.

Home | Main Index | Thread Index | Old Index