Re: object-oriented XML_RPC

First of all, again: Marshall, please respect other programmers. The package is not "messy" just because it uses GLOBALS. It can be discussed if it should use properties instead - but generally GLOBALS is not "messy". If you have concrete suggestions these can be discussed. Not to your mail: As said in the last days: Breaking BC is possible when there is a need for it. Just keep in mind that a new major version number should be used when breaking BC. I haven't used XML_RPC myself - so I can't judge if such a split would be possible / useful. But have you talked to the maintainer of XML_RPC and checked if he maybe would separate his current XML_RPC into two classes? And maybe there could still be a complete XML_RPC- class (keeping BC) that just makes use of the other two classes? Generally I'd love to see extending / changing the existing XML_RPC instead of adding two new packages. And I believe that's the general attitude here as well. About the general structure of the package: I've had a short look at it in CVS. It seems the package doesn't use classes consequently. Maybe the "helper-functions" could also be part of a class and be called as statical methods? Stefan On 20 Aug 2003 at 21:04, Marshall Roch wrote: > I want to use XML_RPC, but it's a very messy looking package (i.e. > $GLOBALs all over the place and repetitive function names like > XML_RPC_ee and XML_RPC_xh). I think it'd be nicer as a class, and > even perhaps as two packages: XML_RPC_Client and XML_RPC_Server. > > I'm going to move it into a class for my own use. Is this something > that would be possible to get into PEAR, or is the lack of > backwards-compatability too big of a problem? How about creating > client and server packages to coexist with the current one instead of > replacing it and breaking backwards-compatability? > > Finally, XML_RPC needs some basic documentation, or at least something > that says it's based off of the usefulinc code (found in the archives, > I haven't confirmed that for myself yet). > > I'm still new to the whole PEAR thing, so maybe my ideas aren't > possible due to the way PEAR works. Let me know whether it's > something I should pursue.