official web-cp RoadMap discussion

[pdrake]

This is the official web-cp RoadMap discussion. Please feel free to share your opinion on what you think is right/wrong about the current RoadMap document. If you are arguing for changes, providing some logical reasoning will give your suggestion a greater chance for adoption. See the RoadMap document on the documents page.

[Gyrbo]

The roadmap is a nice way to see which goals still have to be completed. IF a feature that's only planned for later releases is implemented, will be be put on hold until the release or will it be included anyway?

From what I gather, the major features that still have to be developed for a 0.6 release are:
- virtual users
- modularity of skins/lanuages (would you care to elaborate on this?)
- additional error checking/logging
- Making everything XHTML compliant (1.1 is unrealistic, as I have stated before).
- CSSifying everything. You could also start using the template system for additional things to improve customizability
- knowledge base/ticketing system. I still have something I made a while ago as a standalone script. I'll see if this could somehow be useful.

[pdrake]

If features which are planned for a later release are implemented, I suspect they will be included insofar as they make sense to release in the existing codebase (eg. if someone develops a feature which depends on CVS features/functions we will wait until those dependencies have been released to include it). Yes, you are correct about the features so far which need to be done for 0.6.

Perhaps XHMTL 1.1 is unrealistic but I suppose that will be determined ultimately when I or someone else actually begin the work related to it. Whatever is the most recent standard which can be reasonably used will be the standard adopted for web-cp (doctypes altered to match).

Regarding using the template system for additional things - what things did you have in mind for this?

I've begun on a KB/ticketing system but really only have some basic display / entry functionality and not much else. If you've got code you're willing to donate, feel free to send it to me and I'll see if it can be integrated. Otherwise, I'll continue with my current development.

[Gyrbo]

Allow me to explain why XHTML 1.1 is unrealistic. I've tried writing an XHTML 1.1 validating page, but it simply disallows so many tags and attributes that most older browsers will make a royal mess of your code. target, bgcolor, width, height, collspan, ... all of those can't be used.

As for templates: the tables webcp uses are currently mostly hard coded and only changable via CSS styles. While this is fine most of the time, this can also be templated to allow for more control. The template system can use a fallback template if the requested one isn't available in a specific skin So if the skin doesn't want to do anything fancy, it simply doesn't include the file file. Web://cp will than use the default one. I would personally go a but more cpanel-ish and let each page in the control panel be templated. Using this same fallback it wouldn't require more work for a templater and only slightly more for a page developer. In the end the code would become more managable because data and layout are seperated.

About the KB/ticketing: It's not really anything advanced, but I'm pretty proud of the next/prev page thing (don't ask, I worked on it for several hours). I'll try to explain it with some sample code later (too tired right now), but you can download it here: http://users.pandora.be/gyrbo-be/Minerva-0.1.0.tar.gz I doubt there's much that you can use off the bat in your system, though..

[thalaric]

How about discussing the merge of certain pieces of FlexCP back into Web://cp?

In particular I'm thinking the parser, database abstraction, and updated integrated httpd.

I still haven't decided if it makes more sense to move webcp into flexcp or flexcp into webcp. While webcp is ultimately likely to need everything that is missing in flexcp, at least webcp works today. Some of the finished flexcp stuff is an improvement and would be worthwhile regardless.

[pdrake]

That's a good thought. I do want to port as many pieces of the code as possible since it seems very well thought out. The parser is already in use for skins but it would be great to use it for templates as well. For database abstraction I'm leaning toward ADOdb (with ADOdb-ext for those concerned with speed and willing to recompile) but I could be convinced otherwise. See here for further discussion on that. I'm also interested in the updated httpd server. Can anyone give me a brief overview of the differences between the webcp-httpd daemon and the flexcp one?

[Gyrbo]

Good to see something of you thalaric, it's been a while.

pdrake, thalaric will probably be able to give you a better answer, but I think the main difference is that the flexcp-httpd supports persistent connections. I remember that there were some issues with getting it to work, but AFAIK they were resolved.

[thalaric]

Actually the old httpd accepted persistant connections also. The main difference, aside from bugfixes and general improvements, is that the in the new code the server implimentation is separated out from the protocol itself. This was done because there was lots of discussion on what was the best way to make the server, single process, forking, pre-forking ect. so now it's easy to use one or all of these methods. The server code has been separated out into server.inc.phps and basically creates the looping condition which then just calls a http function to actually act on the data provided by the connection. This also allowed me to use the same server implimentation for the backend socket work as i used for the httpd, even though i used a different protocol for that mechanism.

What's the difference between a template and a skin in Web://CP? You are already using gyrbo's parser class? I'll post my opinion on the DB abstraction.

[Gyrbo]

A template can also be used for non-interface related things (named/apache templates, for example). The skin is a template + images + CSS for the interface.

Web://cp 0.6 already uses my template parser.

[thalaric]

Groovy.