FlexCP + Web-CP integration

jackrabbit · ** Moderator ** Joined: 24 Jun 2004 Posts: 331

I wanted to throw an idea out there for integrating the Web-CP and FlexCP hosting control panel projects. It would go something like this:

1. Web-cp would be renamed to FlexCP. All development work for both projects would be done under the FlexCP banner.
2. FlexCP Version 1.0 would basically be just a final release of Web-CP (if it can ever get to that point). So beginning now the next release of web-cp would actually be FlexCP 0.5.4 then FlexCP 0.6.0 etc., etc.
3. Development work for FlexCP 2.0 would begin with the coding on FlexCP done to date and would continue to be worked on concurrently with FlexCP 0.x-1.0.x. It's first release could be something like 2.0.1.

Basically this idea would be something akin to the direction that the Apache project has taken with its 1.x and 2.0.x projects.

What do people think... Pros ??? Cons ???

Gyrbo · ** Moderator ** Joined: 25 Jun 2004 Posts: 1165

I have a con, there's not really any need. FlexCP is a complete rewrite and that's the reason we picked a different name. Web://cp (or web-cp, or whatever) is a fairly distinct project and I think it would be in the best interest of both projects to keep them under different names.

I am, however, all for the idea to make the developers of both projects co-operate on the same thing. (web-cp until we get a stable release out and a beta for flexcp afterwards).

jackrabbit · ** Moderator ** Joined: 24 Jun 2004 Posts: 331

Doesn't look like anybody is going to bite on the idea. In reality it would just be window dressing anyway. But consider the two projects from an outsiders perspective; work on web-cp basically seems to have stopped in anticipation of the next generation of FlexCP... but FlexCP has hit a roadblock. So basically everything is static. Why would someone want to contribute to a static project?

Anyway, after about a 1/2 dozen patches web-cp works well for me so I have no complaints.

I am still wondering about what we are calling it though... I have kind of switched over to Web-cp. But maybe we should take an official vote on it?

Web-cp
Web-CP
Web://cp
web://cp
Web://CP

???

Gyrbo · ** Moderator ** Joined: 25 Jun 2004 Posts: 1165

My vote would be either web-cp or web://cp. The first indicates that there's a change between the official webcp and the original. This isn't really needed because the official project is dead right now. The current code is a logical continuation of what there was before so I see no problems is using the old name. (Felix has always ben quite liberal in using the name).

I'm not sure if the development of webcp has stopped because of FlexCP, it simply seems to have come to a halt. (For currently unknown reasons)

jackrabbit · ** Moderator ** Joined: 24 Jun 2004 Posts: 331

I think I'm going to rip out the php-httpd server, discard all the old qmail/vqmail stuff, replace the old UW-imap scripts with dovecot compatible code, rework the file upload interface, burn down anything that allows outbound connections to be initiatiated from the control panel, add in a skin, give some of the error codes a new paint job and then release it as a forked project... HackedCP 1.0 !!!

thalaric · Posted: Wed Sep 29, 2004 6:30 pm Post subject:

Do we really need another fork? Why not just fix any problems you have in the official build and submit patches back? I can guarantee any fork you start will end up dead like all the rest. If all the webcp forks would work together we'd be on webcp .7 by now.

I see two reasons for this forking behavior.

1. There isn't strong leadership from the webcp camp
2. no one can ever decide what it is webcp should actually do; so they fork it to fit their own ideas

FlexCP is different because it isn't a hacked up version of webcp. It is written almost completely from scratch to introduce functionality that was never in the original webcp. I don't have any suggestions on how to prevent forks, only observations from watching it for awhile.

jackrabbit · ** Moderator ** Joined: 24 Jun 2004 Posts: 331

The thing is, I've been working on tightening up my servers to not allow outbound tcp-ip connections and this is not compatible with the current webcp. So that represents a major paradigm shift. Then I have about 6-8 hacks that I have to apply everytime I install webcp that i don't think anyone else is really interested in. My "forked" project was a bit tongue in cheek but since webcp for the most part seems static it may be a good time for me to just solidify everything that I need out of it into a single .tar.gzipped package for my own use (and anyone else who's criteria happens to be similar to mine).

thalaric · Posted: Thu Sep 30, 2004 1:08 am Post subject:

If not allowing outbound connections is a good idea then why wouldn't webcp itself benefit from those changes. And what patches do you need to apply that no one else would be interested in? Is it that it's only a problem for certain setups, and if so, are you sure there isn't a better more general way to fix the problem? Or if it's optional behavior couldn't it be in the form of a module?

jackrabbit · ** Moderator ** Joined: 24 Jun 2004 Posts: 331

For the outbound connection functionality... there are traceroute, ping, whois, and dig tools in the control panel that would all be broken by stringent firewall rules. There is also a backup via ftp feature that initiates a connection to an external ftp server. I don't really see the value of this feature but it was added in by the current maintainer of web-cp.net and I doubt he would want to have this particular feature removed.

Here are some examples of hacks I make available:

hostinghacks.net/webcp/htpasswd-hack/hack.txt
hostinghacks.net/webcp/lang-hack/hack.txt
hostinghacks.net/webcp/skin-bluefade/hack.txt

Gyrbo at one point seemed to not like the htpasswd hack. The skin I offered up for merging into webcp on another thread but nobody seems to have picked up on it. I have got some more code for IMAP serving, webalizer and secondary dns syncing if anyone is interested...

thalaric · Posted: Thu Sep 30, 2004 11:56 am Post subject:

I don't see any problem with those patches myself. As far as outbound is concerned that might be more strict than we want overall but easily fixed by some firewall rules.

So I guess what I'm saying is it's easier to fork then get someone to apply your patches, but the latter benefits everyone a lot more. Everyone should try to get their hacks and patches applied as soon as possible. We've lost all kinds of great code because people were working on their own forks that are now abandoned (billing panel anyone? fixed qmail modules?)

jackrabbit · ** Moderator ** Joined: 24 Jun 2004 Posts: 331

I'm certainly more enthusiastic about contributing to a project with a group of developers than running of on my own. So lets go ahead and count off our current developers willing to contribute occasional code updates to the project...

1. j@ckrabbit

Gyrbo · ** Moderator ** Joined: 25 Jun 2004 Posts: 1165

I don't remember having anything against the htpasswd code, do you still remember what it was about?

I'm willing to maintain the sendmail module I've contributed, but I have a lot of other things to work on (FlexCP being a higher priority than web://cp). I'll fix bugs if they arrise, but I don't really have plans to contribute more major code.

jackrabbit · ** Moderator ** Joined: 24 Jun 2004 Posts: 331

you were just recommending a better way to do it so that multiple users could get added into the .htpasswd file. the current setup only allows for a single user and any new information overwrites the old file. just a common sense line of thinking really. I didn't take it personally. it's pretty sloppy code overall ( doesn't do any error checking or stuff like that ).

I'm thinking about spending the next couple of months totally ripping the guts out of webcp and then trying to put it back together. kind of like that one time I took my motercycle engine apart and then pieced it back together. it was a little disconcerting having those extra bolts left over at the end of it all but it never fell apart or anything like that. Laughing

But maybe after it all i'll have developed enough skills to help out with the FlexCP project.

Anyway, i'm heading out to Hawaii for two weeks of fun in the sun. I'll check in on you guys when I get back.

Gyrbo · ** Moderator ** Joined: 25 Jun 2004 Posts: 1165

I'd be more concearned if you got some bolts left from web://cp. It uses tape and glue most of the time...

i'll see if I can put some work in FlexCP from my side, but I'm currently developing a paid project. Sigh, I'm already missing templated classes in PHP, oh well.

pdrake · ** Site Admin ** Joined: 24 Jun 2004 Posts: 1053

jackrabbit, didn't we discuss the outbound IP connections previously? I think that what we determined was that it would be good for the system administrator to have an option in the control panel to turn on and off the network tools (eg. ping, tracert, etc) and turn on and off the remote backups feature. Wasn't that what we had decided would allow the control panel to accomodate the widest range of needs? Maybe I'm not remembering correctly. From reviewing the 3 hacks that you posted, I think they would make good additions to web-cp if you are offering the code to be included.