Re: [Zope-dev] ZPatterns example update....
Steve Spicklemire wrote:
Hi Folks,
The Dumb ZPatterns example is updated. Now there is some more realistic object referencing going on, borrowing of code snippets between Specialists and suchlike. There is also more in the way of reasonable documentation, though everything is in flux, and it still doesn't resemble a truly completed product. I just feel the need to get things finished enough that I can stop thinking about them for a while. ;-) In particular there the ToDos now hold references to Doer and Deliverable, and no 'lists' are maintained. One thing I need to do is to have these references automatically 'fixed' when a 'referred to' entity is removed or modified in such a way that the link should be broken.... that's for the next version. If I'm not careful.. it won't be a 'simple' example anymore. There's got to be a line here somewhere (in the sand?).
Nice work, Steve. Removing id lists looks good. I started building objects using getXForY(y_id), but the last one I made used an id list following the Dumb Example. I've just changed it to use getXForY, and it's much cleaner, plus it makes adding X from Y/editInstanceForm easier. I like the idea of naming all UI methods with '_html' - now, if you find yourself referencing a method which doesn't end in '_html' in your html code, you know you need to add an interface method. I just need to decide if I like it enough to change all my existing code... Some thoughts about the broken links handling problem: If an object depends on the existence of another (for example, if you wanted a ToDo to be tightly linked to a Deliverable) then it should be deleted when the Deliverable is deleted, right? You'd have WHEN OBJECT DELETED CALL ToDos.deleteInstances(myToDoIDs) in the Deliverable SkinScript, and you would never have ToDos floating around without a Deliverable. As for the Deliverable changing and invalidating the link, I think it would be solved if you used immutable ids for all objects, and stored the Deliverable title in a separate property. In case of weaker links, such as between ToDo and Doer, I guess it should be ToDo's responsibility to return None if the Doer referenced by doerID doesn't exist - treating it the same way as the case where no doer has been assigned. Maybe with this SkinScript: WITH Doers.getItem(self.doerID) CALCULATE self.doerID=RESULT.id or '' ? But I'm not really sure about this... I think you've managed very well to stay on the right side of the line in the sand... the problem is that complex real world applications have a lot of stuff on the other side of the line, and the challenge is to pull that stuff over the line. My personal problem is figuring out if the stuff I have on the wrong side of the line is necessitated by the complexity of the application, or if I'm just making things unnecessarily complicated... Itai -- Itai Tavor "Je sautille, donc je suis." C3Works itai@c3works.com - Kermit the Frog "If you haven't got your health, you haven't got anything"
Hi Itai,
"Itai" == Itai Tavor <itai@optusnet.com.au> writes: Itai> Steve Spicklemire wrote:
>> Hi Folks, >> >> The Dumb ZPatterns example is updated. Now there is some more >> blah blah blah.... Itai> Nice work, Steve. Thanks! Itai> Removing id lists looks good. I started building objects Itai> using getXForY(y_id), but the last one I made used an id Itai> list following the Dumb Example. I've just changed it to use Itai> getXForY, and it's much cleaner, plus it makes adding X from Itai> Y/editInstanceForm easier. Yup.. this is working nicely for me. Itai> I like the idea of naming all UI methods with '_html' - now, Itai> if you find yourself referencing a method which doesn't end Itai> in '_html' in your html code, you know you need to add an Itai> interface method. I just need to decide if I like it enough Itai> to change all my existing code... Right... I was on the lookout for something natural, and sorta obvious. I liked the fact that there are already examples of 'standard' interface methods that have '_html', so I wouldn't have to change *them*... sometimes laziness can be inspiring. ;-) Itai> Some thoughts about the broken links handling problem: If an Itai> object depends on the existence of another (for example, if Itai> you wanted a ToDo to be tightly linked to a Deliverable) Itai> then it should be deleted when the Deliverable is deleted, Itai> right? You'd have WHEN OBJECT DELETED CALL Itai> ToDos.deleteInstances(myToDoIDs) in the Deliverable Itai> SkinScript, and you would never have ToDos floating around Itai> without a Deliverable. As for the Deliverable changing and Itai> invalidating the link, I think it would be solved if you Itai> used immutable ids for all objects, and stored the Itai> Deliverable title in a separate property. In case of weaker Itai> links, such as between ToDo and Doer, I guess it should be Itai> ToDo's responsibility to return None if the Doer referenced Itai> by doerID doesn't exist - treating it the same way as the Itai> case where no doer has been assigned. Yes.. I like this. Itai> Maybe with this Itai> SkinScript: WITH Doers.getItem(self.doerID) CALCULATE Itai> self.doerID=RESULT.id or '' ? But I'm not really sure about Itai> this... Ahh.. I think my brain just imploded in a recursive death spiral, but I get intent of the idea. This would work I think: WITH Doers.getItem(self.doerID) COMPUTE self.myDoer=RESULT or Doers.getItem('doNothing') where there is a default 'doer' named 'doNothing' in one of the Doers Racks. This guy would 'fill in' for the ToDo's doer when no 'real' doer can be found... I like that. ;-) Itai> I think you've managed very well to stay on the right side Itai> of the line in the sand... the problem is that complex real Itai> world applications have a lot of stuff on the other side of Itai> the line, and the challenge is to pull that stuff over the Itai> line. My personal problem is figuring out if the stuff I Itai> have on the wrong side of the line is necessitated by the Itai> complexity of the application, or if I'm just making things Itai> unnecessarily complicated... Thanks... don't want to mix metaphors, but I'm fearing that the line in the sand may be on a slippery slope ;-). Seriously though. It would be really easy to make this 'dumb' example so complex that a beginner in ZPatterns couldn't make it through without brain damage.... but never fear.. I've already squandered all my free time for the next three months on this... so it will very likely stay stuck where it is for some time now... I hope that some ZPatterns beginners find it useful as a starting place. take care, -steve
Hi Steve,
Steve Spicklemire wrote:
"Itai" == Itai Tavor <itai@optusnet.com.au> writes:
Itai> Maybe with this Itai> SkinScript: WITH Doers.getItem(self.doerID) CALCULATE Itai> self.doerID=RESULT.id or '' ? But I'm not really sure about Itai> this...
Ahh.. I think my brain just imploded in a recursive death spiral, but I get intent of the idea. This would work I think:
WITH Doers.getItem(self.doerID) COMPUTE self.myDoer=RESULT or Doers.getItem('doNothing')
where there is a default 'doer' named 'doNothing' in one of the Doers Racks. This guy would 'fill in' for the ToDo's doer when no 'real' doer can be found... I like that. ;-)
Yeah, this looks good... whether it's appropriate or not depends on how you expect the case where there's no Doer to behave - if you needed to know explicitly that a Doer doesn't exist, returning a 'nothing' doer could confuse things. BTW, I can't see the recursion in my script... but maybe my brain is still in vacation mode :)
Itai> I think you've managed very well to stay on the right side Itai> of the line in the sand... the problem is that complex real Itai> world applications have a lot of stuff on the other side of Itai> the line, and the challenge is to pull that stuff over the Itai> line. My personal problem is figuring out if the stuff I Itai> have on the wrong side of the line is necessitated by the Itai> complexity of the application, or if I'm just making things Itai> unnecessarily complicated...
Thanks... don't want to mix metaphors, but I'm fearing that the line in the sand may be on a slippery slope ;-). Seriously though. It would be really easy to make this 'dumb' example so complex that a beginner in ZPatterns couldn't make it through without brain damage.... but never fear.. I've already squandered all my free time for the next three months on this... so it will very likely stay stuck where it is for some time now... I hope that some ZPatterns beginners find it useful as a starting place.
I'm sure this example will be useful for beginners... it answers a lot of questions that it took me weeks to figure out. And it worked well as a test bed for various implementations of object connections. Shame about your free time :( Itai -- Itai Tavor "Je sautille, donc je suis." C3Works itai@c3works.com - Kermit the Frog "If you haven't got your health, you haven't got anything"
At 03:08 PM 1/2/01 +1100, Itai Tavor wrote:
Hi Steve,
Steve Spicklemire wrote:
> "Itai" == Itai Tavor <itai@optusnet.com.au> writes:
Itai> Maybe with this Itai> SkinScript: WITH Doers.getItem(self.doerID) CALCULATE Itai> self.doerID=RESULT.id or '' ? But I'm not really sure about Itai> this...
Ahh.. I think my brain just imploded in a recursive death spiral, but I get intent of the idea. This would work I think:
WITH Doers.getItem(self.doerID) COMPUTE self.myDoer=RESULT or Doers.getItem('doNothing')
where there is a default 'doer' named 'doNothing' in one of the Doers Racks. This guy would 'fill in' for the ToDo's doer when no 'real' doer can be found... I like that. ;-)
Yeah, this looks good... whether it's appropriate or not depends on how you expect the case where there's no Doer to behave - if you needed to know explicitly that a Doer doesn't exist, returning a 'nothing' doer could confuse things. BTW, I can't see the recursion in my script... but maybe my brain is still in vacation mode :)
Your script uses an attribute (doerID) in the WITH clause that is provided by the COMPUTE clause. This is infinitely recursive in theory. In practice, it will result in the non-existence of the doerID attribute, as ZPatterns treats recursive attribute references as non-existence of the attribute. The WITH clause will fail with an AttributeError.
participants (3)
-
Itai Tavor -
Phillip J. Eby -
Steve Spicklemire