[cl-dwim-devel] Adding the concept of a natural key to perec

[Attila Lendvai]

> In either case, it saddens me.  The dwim libraries are some of the
> most ambitious, innovative, and substantial in the body of
> contemporary common-lisp.  (IMHO). On the bright side I'm probably
> years from fully understanding all of it so at least you've left a
> prodigious bone for others such as myself to gnaw on. :).


nice words, thanks! i hope you can learn from it or make money with it
in a more direct way!

as of the future, it's not like we gave up programming forever... ;)

but if you ask me, these days i more often than not slide into a
regression whenever i think about anything interesting in programming:
i always end up thinking about how to fix the environment, and then
write 5 lines to implement the random cool feature i'm pondering
about, instead of reinventing a broken wheel a thousandth times...

alan kay's vpri.org and the fonc project seem rather interesting, but
psychology and stuff around it seems to capture my brain even more
these days. so much i should have learned much sooner...


> I do hope, though, that its not the end of your work extending and
> refining these libraries -- I think they contribute to the ongoing
> viability of CL as a modern platform.  In particular it would be
> wonderful if the web meta app meta server were to be updated and
> fleshed out a bit more someday.  Maybe not to the standard set by
> perec (which is rock solid, in my experience) but maybe a bit
> closer...  I haven't tried the sencha version though -- is that
> operational?  A more recent JS framework like sencha/extjs seems
> like it might offer a substantial upgrade to the capability in terms
> of UI.


the sencha stuff is rather half done.

the major issue that bothers me is that the web-server/presentation
split is not done properly yet. there are a few kludges between
them. and the tree of entry-points in the web-server part could also
be cleaner, but i guess i lack some insight to simplify it. there's an
abstraction somewhere there that can be used to declaratively describe
a tree of handlers of all kinds, and then use the statically available
conditions to generate/compile a fast dispatcher.

and the components in the presentation project are also halfway in a
refactor.

oh, and i really don't like the implementation of quasi-quote...

ahh, that never-ending dissatisfaction with one's own code... :)

</rant>

-- 
• attila lendvai
• PGP: 963F 5D5F 45C7 DFCD 0A39
--
“That language is an instrument of human reason, and not merely a
medium for the expression of thought, is a truth generally admitted.”
	— George Boole