- Subject: Re: Fixes to "recent" and "tree" modes
- From: g.milde@xxxxxx
- Date: Tue, 3 Feb 2004 08:20:04 +0100
Paul Boekholt wrote:
> On Fri, Jan 30, 2004 at 11:42:14AM -0800, Dave Kuhlman wrote:
> > In recent.sl, I changed:
> >
> > autoload("what_line_if_wide", "bufutils");
> >
> > to:
> >
> > autoload("what_line_if_wide", "sl_utils");
Thanks for the report, it is fixed now.
> Apparently Günter changed his mind about where to put what_line_if_wide.
> The best way to manage the autoloads for Günter's *utils is IMHO to not
> write autoloads in the modes that use them, instead the user should use
> make_ini.sl to make his own autoloads.
Of course, using make_ini (also on jedmodes.sf.net) makes life easier for
the mode developer (as (s)he doesnot need to think about the autoloads. For
modes intended for widespread use, I still prefer the explicit way -> less
dependencies, more transparent. (The backside is, that as I use the make_ini
method myself, I sometimes miss to spot the bugs..)
Optimizing the modes "layout" is tricky. My priorities are:
- Fast startup - load only what is needed, other stuff can wait.
- code recycling - reuse helper functions whenever possible
- Modularity - makes the above points easier/possible
For the decision what mode a function should belong, there are sometimes
contradictory side-constraints:
- Keep stuff that belongs together in one file (data-utils, buffer-utils,
string-utils, ...)
- Backwards compatibility
- Let user-interface modes depend on as few as possible other modes:
sl_utils - most basic functions needed by almost any other mode
(so, e.g. what_line_if_wide is needed by navigate.sl which
autoloads a bunch of sl_utils functions but no other one from bufutils)
Günter
--
G.Milde at web.de
--------------------------
To unsubscribe send email to <jed-users-request@xxxxxxxxxxx> with
the word "unsubscribe" in the message body.
Need help? Email <jed-users-owner@xxxxxxxxxxx>.
[2004 date index]
[2004 thread index]
[Thread Prev] [Thread Next]
[Date Prev] [Date Next]