From: Taku Nakajima Date: 2002-12-30T21:20:32+09:00 Subject: Re: Another take on RDoc's templates At Sat, 28 Dec 2002 17:16:45 +0900, Bruce Williams wrote: > > On Saturday 28 December 2002 02:31 am, Tom Sawyer wrote: > | currently i use eRuby. I would have liked to use Amrita, but it seems the > | attribute templating is still considered experimental. > > If you're talking about attribute expansion; yes it is still 'experimental' > (as of 1.0), but I speak from experience when I say it is definitely working. > I use amrita for the vast majority of my web projects (in the past I used > eruby) and attribute expansion has never been a problem. I'm guessing it is > only considered "experimental" due to the fact its implementation may be > changed sometime in the future (I defer to Nakajima-san for further details > and/or clarification). I agree with you. I think "attribute expansion" is working well . amrita's experimental features are experimental partly because they are not tested enough. But main reason is it's easy to add a feature but difficult to remove one once it was released and get users. > | i'm curious as to the relative performace comparisons between the various > | templating options. anyone have any experience to draw upon for this? > | > | -transami > > Eruby/erb is decidedly faster than amrita due to the way data is applied to > the templates (amrita actually understands the template 'context' when the > data is applied, enabling it to apply sequences repetatively, etc-- this > "smart" templating comes at an obvious perfomance cost)-- other templating > libraries blindly insert data, IIRC. Amrita is, of course, an excellent > choice for static page creation, but I would caution using it in larger, > heavy bandwidth projects where templates would be expanded dynamically. > Speed optimization is definitely one of the chief areas of concern in Amrita > development; I'm sure the situation will become more amenable in the future > as the project matures. Recently a good idea of new compiler architecture came to me . I think the new version of amrita will produce a Ruby code as fast as a plain Ruby. And it will also generate a C code (extention module) that runs faster than a Ruby equivalent. -- Taku Nakajima http://www.walrus-ruby.org/amrita/