>Well the parts in Avalon CVS are the "base" of my
>cron job server that did most of what you wanted ;)
>
>From memory the other blocks I created were
>
>PersistentJobStore - stores job definitions and scheduling
>JobDispatcher - ie used the job defintions to create list of ant tasks and
>execute them
>SubjectManager - ie had the subjects/principles and their associated
>permissions
>CronScheduler - Manager all the rest of the components
>
>Except for SubjectManager (which was an absolute PITA) all the rest are
>relatively easy to implement. We had both a DB based PersistentJobStore and
>a file based PersistentJobStore.
>
>The only reason it is not free software/opensource at the moment is because
>(a) it relies on a very very ugly fork of ant code (enough that I am
>ashamed to call it my own ;])
>(b) I don't have sole copyright on most of it ;(
>
>As soon as the task engine for ant2.0 is complete I will rebuild it again
>though ;) It is largely comparable to fluxs server but it did not have some
>of the more complex stuff (they had automatic notification/events for
>certain cron events IIRC). Any it will be hosted in Avalon/Phoenix ;)
>
>Cheers,
>
>Pete
I'll look more closely at what Avalon offers today. (I'll also try to give
Turbine's Scheduler a close look, and ANT).
Frankly I'm a little disappointed that all of this "furor" got restarted
about the whole commons business - but I guess it just proves that things
aren't settled down yet... I'm still not sure what it is supposed to be.
There are some very fine lines to walk when trying to build code that is
re-usable, and also clean and straight forward, and also tries to re-use
some code itself. For instance, I immediately cringe at thinking the job
scheduler should be dependent on ANT - I'd like something that is as
dependent on as few other things as possible... However, if ANT 2 has a
great task engine, then it would be a shame (waste of time) to reproduce
another. However, if ANT's task engine isn't built in a way that fits
naturally with the rest of the Job Scheduler I'd like to have, then why
shoe-horn it in?
Another big problem with relying on too many outside products within your
own can be exemplified with the following example: Weblogic uses ANTLR to
parse JSPs. One of my companies products also uses ANTLR - but the later
versions of ANTLR. Weblogic uses an older (non-binary-compatible) version of
ANTLR, and therefore our product works fine on every app server but
weblogic.
I also agree with the poster on this thread who said that libraries/products
should not be tied to a framework, but rather the framework should
incorporate them. I'm positive that if the Job Scheduler is tied to Avalon
or Turbine or XXXX framework, many people (including myself) will not be
very happy about using it within a product that already has it's own (and
very different) framework.
Don't get me wrong - I actually very much want to re-use code, but I think
when you do so, you need to be very careful and prudent about which things
you include, and how many of them. It could help a lot if projects weren't
packaged as one big jar with interdependencies between all of the packages
within the jar. For example, if we wanted to use ANT's task engine it would
be nice if there was an "ant_tasks.jar" not just an "ant.jar".
I initially proposed creating a open source Job Scheduling project because I
was unaware of one (didn't know there was one in Avalon, and I assume the
one in Turbine currently only works within Turbine) - and I'm sure it's
something that MANY people could use and appreciate having. I'd really like
to leverage existing code (it looks like we have at least 4 sources of
existing job scheduling code - mine, Avalon's, Turbine's, and ANT's).
I need to get started on something right away, but would really like it to
be developed as part of a community, and have the aid others in developing
it. How soon will ANT 2.0 be complete? I really would hate to build this
thing myself, and then watch you (Pete) build your own a few weeks later
(you seem fairly determined to rebuild yours once ANT 2 is complete).
Where do I go from here? -- I guess I'll study up on ANT, Avalon and
Turbine, and then write up a proposal. I'll be VERY happy to have the
project fall somewhere (I'm open to anywhere) under Jakarta, but if there's
too much bickering I may have to go my own way :(
At least there seems to be good support for the second project I proposed
(the process automation engine), without the headache of pre-existing
projects that overlap. Bip Thelin and I will be submitting a proposal in the
near future. -- Unfortunately, this project will almost certainly require a
Job Scheduler, and the arguments will have to start again!!!!!!!!!!!
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
