My preference is for option 1, for a couple of reasons: a) Please for the love of all that is holy, let's not change the plan AGAIN! It has been said and resaid on this ML that "Genshi will be the default in TG2". Even as similar as Genshi is to Kid, it caused a pretty big uproar. The speed argument did not convince people then. The big advantage to Genshi that actually convinced people (IMO) was the improved error reporting, and the fact that Kid was only semi-maintained. Switching to Mako as default I think will tick off a fair amount of people.
b) As Diez said, "Developer time is more precious than CPU-cycles." I find xml-based templating drastically cuts down on time of development, and makes it easier to coordinate with a separate developer. The developer time we save by not changing over can be used to work on features that were very exciting and attractive to the user community when TG 0.8 was launched, and have been largely unmaintained since. (Catwalk, Model Designer, etc.) c) If I were to use something besides Genshi/Kid, I would spend way more time validating pages, etc. Ick. :) I don't think TG was envisioned as the "Framework for making fast apps". IMO it was designed to reduce grunt-work and repetitiveness for the developer. Let's stick to that philosophy. Kevin Horn On Dec 13, 2007 3:58 AM, Diez B. Roggisch <[EMAIL PROTECTED]> wrote: > > On Thursday 13 December 2007 05:40:01 Mark Ramm wrote: > > So, I like Genshi, and it's definitely more "backwards compatable" > > with tg1.x. So, we definitely need to support it. But genshi will > > always be slower than Mako. And mako has lots of page fragment > > caching features which are very nice for high traffic sites. > > > > People who have lots of traffic, or want the lowest possible latency, > > will probably always benefit from using mako, and if we start building > > reusable turbogears components, it would make sense to use mako since > > that would make them re-usable for the widest variety of people. > > > > As I see it we have a few options: > > > > 1) Make Gehsh the default, use Genshi in components, and leave the > > speed daemons to do stuff on their own. > > 2) make mako the default, use mako to build components, and break > > backwards compatibility > > 3) try to get component developers to provide both Genshi and Mako > > versions (ToscaWidgets does this to some extent IIRC). > > > > I'm concerned that option 1 limits the performace/scalability of tg2, > > and option 2 is too big of a change from tg1. And I'm not sure how > > much work 3 is, but I'm guessing that it's not insignificant, so it > > will limit the number of components that get developed. > > > > What do you all think? > > IMHO what TG needs is features. People on the ML often refuse to dive into > intricacies of e.g. AJAX, remote forms and the like - they want solutions > to > their problems. > > So I think we should use what helps us to whip out components and features > as > fast and completetely as possible. Which IMHO is genshi. Developer time > is > more precious than CPU-cycles. > > Just the other day I read about youtube and their strategy to enable high > perfomance, while using python. And I think that points into the same > direction: raw processing speed never wins by itself - you need to > carefully > tune you application, sure. But competition in the market is about > features - > you can have the fastest C-backend to server videos - if it sucks > featurewise, nobody will use it. > > Diez > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "TurboGears Trunk" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/turbogears-trunk?hl=en -~----------~----~----~----~------~----~------~--~---
