On Sun, 2004-02-08 at 13:32, Jason Maas wrote: > Hi Derek, > > On Sun, 8 Feb 2004, Derek Neighbors wrote: > > >It is disappointing to hear people rule out GNUe when as best I can tell > >they never dialoged with any GNUe developers. > > Thanks for posting, it's good to hear your report that there is life in > the GNUe project! I'm someone else who looked at the GNUe website and > decided that there didn't seem to be much momentum and activity. You do > make a very good point that people shouldn't write off GNUe without > contacting the developers. However, I'll tell you why I didn't think much > was happening -> no mailing list activity! One of the first things I do > when checking out a project is look for activity on the development > mailing list(s). There's practically none on 'gnue-dev'. You should probably look at CVS commit data to see life of a project more than mail list traffic (IMHO). CVS is taking several commits a day. > If someone happens to look at the page about the IRC channel the > developers use, *then* they see the text about IRC being the main > communication method between devlopers, *unlike* other projects. > Personally, I don't like IRC as a main developer communication method > because it's hard for new people to see the life of the project on its > website, and its harder to track down previously discussed issues and > threads of discussion. Plus, the IRC logs are on a different website > which appears to be down at the moment, so I'm not even sure what they > have and how useful it is. We use Kernel Traffic Kernel Cousins[0] to do something similar to a mailing list archive for IRC. The big problem is we are much more like an office of developers than a distributed development team. This has decided advantages in that we get answers immediately as if we were in the same office. It has disadvantages in that people that prefer delayed communication like mailing lists feel the project is dead. We are not opposed to mailing lists and will respond to mails sent to them, but why force developers to communicate in a way that is not natural to them? > So maybe you guys should try to bite the bullet and move your discussions > to mailing lists like most projects, or at least put a more prominent > message on your website saying that the mailing lists aren't really used > and IRC is the way to contact anyone about the project. I will pass the information forward. There definitely needs to be better indicators as to the life of the project. > >[...] > >That said we had an appserver that went strongly down the > >object/relational mapper. I think for real business application > >development this is a mistake. > > Why do you think it's a mistake? I'd like to hear more about what you > guys have learned from your GNUe AppServer experience. This is a longer response that I don't have full time for as I spent a good deal of time last night (or recently) talking to someone about this. The two big reasons. Objects are great for somethings, but are tremendously poor at others. In the same way relational queries are great for some things, but poor for others. Another practical problem is in order to optimize properly object to relational mapping you end up with relational tables that are so hosed up they don't become very practical for querying relationally. At which point you have to ask why bother? Why not just use an Object Database. The other reason is why objects are very efficient for some operations they are extremely (and I mean extremely) resource hungry for others. I have not only done lots of object/relational mapper stuff within GNUe and outside of GNUe I have had to develop with more than one proprietary true object databases. I am not against the concepts, but in practice the theory is less than perfect. It is a matter of finding the sweet spot. -- Derek Neighbors GNU Enterprise http://www.gnuenterprise.org derek@gnue.org Was I helpful? Let others know: http://svcs.affero.net/rm.php?r=dneighbo
Attachment:
signature.asc
Description: This is a digitally signed message part