Re: This Wiki Implementation

["John Belmonte" <john@...>]

Here are my random thoughts in response to this thread.

It's still the case that languages like Python and Perl are superior
for completing the average random programming task in less time than
Lua, given their large and mature base of support libraries.  For
example, recently I had to add spam detection support to the Lua wiki.
 Despite my dislike of Perl (the wiki's language), it was certainly
convenient to use an already-written module for communicating with a
popular spam filtering service.

Nevertheless, give the right circumstances I'm not ruling out a Lua
implementation.  In fact, a few months ago I tried to reach Daniel
Silverstone about migrating to his Aranha Wiki engine.  Perhaps my
message ended up in his spam box...  Anyway, the reasons I would
consider that wiki are:  1) it painstakingly emulates the current Lua
wiki markup;  2) it preserves idioms of the current wiki (e.g.
clicking on title yields referring pages);  3) its interface and
usability are very simple (e.g. simple preferences page, registration
not required for editing);  and 4) it is strict about web standards.
Work was still needed to add spam support though, related to my next
point.  If a wiki can do all these, it comes down to the advantage of
trading a big mess of Perl (mainly due to a ton of features we don't
use in the current engine) for a little Lua.  I'm not interested in
trading a big mess of Perl for a big mess of Lua.

Functionality to automatically deal with spam is absolutely critical
for a popular, no-registration wiki such as as lua-users.  Prior to
having a workable solution in place this really was a detriment to my
quality of life, to the point where I dreaded visiting the wiki each
day.  It's not easy to get this stuff right, and any replacement wiki
needs to preserve what exists now (e.g. chaining of filter services,
retry on service failure, white-listing of some users by cookie in
order to train filter services).

The Lua wiki engine is old, but its limitations are well known and we
have about 7 years of evidence that they aren't show-stoppers.  I'm
not much bothered by the site not using a Lua wiki-- aren't wiki
engines a commodity by now?


Regards,
--John


On Feb 14, 2008 12:32 PM, Petite Abeille <petite.abeille@gmail.com> wrote:
> Re: http://lua-users.org/wiki/ThisWikiImplementation
>
> John Belmonte lists a couple of points regarding the current wiki
> implementation (Perl + UseModWiki) powering lua-users.org as well as
> any potential replacement.
>
> In no particular order:
>
> * The markup and to a certain extent the interface is what really
> defines a wiki system
> * CSS
> * 90% of a wiki implementation is spent on details such as database
> access, edit conflict resolution, page history management, security
> and robustness, etc
> * languages such as Python and PHP may be better suited for such
> implementations due to their extensive web and system support libraries
>
> Now that there is a least two rather full-featured wiki
> implementations in Lua, namely Sputnik and Nanoki, would it be
> worthwhile to ponder updating the current setup in favor of a Lua
> implementation?
>
> http://sputnik.freewisdom.org/
> http://alt.textdrive.com/nanoki/
>
> Thoughts?
>
> Cheers,
>
> PA.
>