[Koha-devel] RFC: proposing change to Koha's plugins system

[Paul A]

At 10:16 PM 8/21/2014 +0000, Indranil Das Gupta wrote:
>Hi all,
>
>I've been trying to use the KPZ structure in some of my projects. In
>general, I've found it rather useful, with a few modifications.
>
>I would like to propose the following changes as
>http://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=12805
>
>lets have your feedback on this.

With a view of maintaining core stability for Koha, all 'enhancements' 
(e.g. plug-ins) should be optional and not affect speed, size and cpu 
cycles of basic functionality.  I think that this is what you are 
proposing... if I'm mistaken, please advise.

I've been relatively quiet for some time about my intrinsic belief that a 
two year cycle of updating Koha (along the same 2-year lines of most OS's) 
should be the norm. "If it ain't broke, don't fix it."  Enhancements (not 
security updates) should be for the end-user to decide upon, and most 
libraries do not have the personnel | time | budget to get involved every 
couple of months.

I recognize that the paid consultants may have an interest -- and I have 
absolutely nothing against that; without them a huge amount of "Koha devel" 
would not take place.  However, we are getting ready to upgrade our servers 
from 2012 to 2014 standards, and I've been looking at Koha 3.14 and 
3.16.  We are *not* a lending library -- just a big research centre -- and 
quite frankly I can find no totally compelling reason to upgrade from 3.8 
(handheld browsers might be the exception.)  Again, if I'm mistaken, please 
advise.

Please don't misunderstand me.  I'm all in favour of "improvments", but 
when a quick analysis of one of my OPAC pages come up with:

Combine external CSS (6)
Combine external JavaScript (11)
Leverage browser caching (23)
Leverage proxy caching (5)
Minimize cookie size
Serve static content from a cookieless domain (11)
Specify image dimensions (1)
Optimize the order of styles and scripts (2)
Remove unused CSS rules (1335)
Use normal CSS property names instead of vendor-prefixed ones (3)

it's probably better to look at the 1,335 "unused CSS rules"... (bootstrap? 
doesn't do away with yahoo? does zebra depend on yahoo?)

End rant. I promote Koha wherever I can, but in my job I've got to stay 
practical. Over the last four years we've put in place a marvelous OPAC 
meeting our users requirements.  Please keep up the good work, but please 
also maintain core stability.  Compare with Debian (lenny, squeeze, wheezy, 
2 year cycle); Ubuntu (lynx, pangolin, tahr, 2 year cycle); Apache (2.2 ==> 
2.4, eight/ten? year cycle); MySQL (fairly minor in the 5. series, no 
urgency); Perl (ditto); etc...

Best -- Paul