ruote tmp/log_2012-05-07.html

2012-05-07 02:56:44 utc Mugatu I *think* this may be something new, but has anyone noticed that the ruote-flo chart as seen in ruote-kit only displays a small corner of the flow?
2012-05-07 03:16:00 utc jmettraux Mugatu: hello, do you have more details? OS and browser versions?
2012-05-07 03:17:02 utc Mugatu jmettraux:sure thing - Chrome 19 on MacOS
2012-05-07 03:17:16 utc Mugatu running from git master for both ruote and ruote-kit
2012-05-07 03:17:46 utc jmettraux OK, trying Chrome 18.0.1025.168 on snoleo immediately
2012-05-07 03:19:40 utc jmettraux ruote-fluo master renders fine
2012-05-07 03:19:43 utc jmettraux trying ruote-kit now
2012-05-07 03:20:27 utc jmettraux ruote-kit renders fine as well
2012-05-07 03:20:43 utc jmettraux maybe your Chrome 19 is too "edge"
2012-05-07 03:21:25 utc Mugatu hah, could be, just for fun, here's what it looks like for me: https://www.evernote.com/shard/s3/sh/62bbe9f2-c992-40e7-99bd-09016b838b58/f6c4dc06782dbb8908f8ff987d14556f
2012-05-07 03:22:44 utc Mugatu I'm so out of touch with firefox, I don't even know if it should support canvas properly or not, I don't see any image at all there (version 11.0) -- but just checked safari and all is well
2012-05-07 03:22:57 utc Mugatu sorry for the false alarm, must be browser-related
2012-05-07 03:23:05 utc jmettraux no worries
2012-05-07 03:23:19 utc jmettraux there's probably something showing up in your javascript console
2012-05-07 03:23:24 utc jmettraux like the resize bit failing
2012-05-07 03:23:56 utc jmettraux on the other hand, I started work on an SVG version of ruote-fluo, it might interest you: https://github.com/jmettraux/ruote-fluo/tree/svg
2012-05-07 03:23:58 utc Mugatu ah hah: "Uncaught TypeError: Cannot set property 'flow' of undefined"
2012-05-07 03:24:09 utc Mugatu I wonder if that could be the root
2012-05-07 03:24:46 utc Mugatu that's from the renderFlow function in ruote-flo.js
2012-05-07 03:25:04 utc Mugatu svg would be awesome too :)
2012-05-07 03:25:23 utc jmettraux what line?
2012-05-07 03:25:34 utc Mugatu 879 in my version
2012-05-07 03:25:40 utc Mugatu context.canvas.flow = flow;
2012-05-07 03:27:01 utc jmettraux so context canvas is "undefined"
2012-05-07 03:27:07 utc Mugatu yep
2012-05-07 03:27:18 utc Mugatu just running through the JS debugger, actually context at this point is actually a HTMLCanvasElement
2012-05-07 03:29:56 utc jmettraux I have to go for lunch now, do you have time to fill an issue report for it?
2012-05-07 03:30:00 utc jmettraux well I guess it's https://github.com/jmettraux/ruote-fluo/issues/2
2012-05-07 03:30:16 utc jmettraux if you could add details into that issue, that'd be great
2012-05-07 03:30:21 utc Mugatu sure thing, thanks!
2012-05-07 03:30:31 utc jmettraux thanks!
2012-05-07 04:30:56 utc Mugatu Could someone help me validate this approach (This has to be the most basic use case :)) https://gist.github.com/f1f79111565010d3490a Effectively, I need to update a status workitem field externally, and not continue with the workflow until it matches a specified gating value.
2012-05-07 04:31:38 utc Mugatu The workflow actually executes OK, but I'm left with dangling workitems and expressions, breaking the ruote-kit interface (not my primary issue, but I'm wondering if it points to a larger problem with the approach)
2012-05-07 04:33:06 utc jmettraux so those workitems are stored in a storage participant?
2012-05-07 04:33:54 utc Mugatu yes - form_submitter is just falling to the catchall
2012-05-07 04:34:08 utc Mugatu (using the redis storage engine, although I don't think that should matter)
2012-05-07 04:34:22 utc jmettraux you're right, it doesn't matter
2012-05-07 04:34:54 utc jmettraux calling receive on the engine is making the flow resume, but it doesn't tell the storage participant to delete the workitems for its "bucket"
2012-05-07 04:35:15 utc jmettraux you could replace it with RuoteKit.storage_participant.proceed(wi)
2012-05-07 04:35:24 utc jmettraux that should do it
2012-05-07 04:36:42 utc Mugatu thanks, giving it a spin now
2012-05-07 04:41:35 utc Mugatu getting there :) workitems are all cleared now, although I was left with 4 instances of the 'notifier' expression (the notifier is a custom participant), all with a state of 'cancelling'
2012-05-07 04:41:48 utc Mugatu Not sure if I just need to implement on_cancel, doing that now
2012-05-07 04:42:43 utc jmettraux yes, without the on_cancel, the process will complain about it missing and you'll have an error
2012-05-07 04:44:28 utc Mugatu hmm, I don't think that's it (I've added that and tried again) -- I don't have any ruote errors reported, except for a breakage in the ruote-kit process list ("NoMethodError - undefined method `ancestor?' for nil:NilClass:"), which I assume is because these expressions don't have the correct parent structure
2012-05-07 04:46:00 utc Mugatu (The notifier does call 'reply' from within its on_workitem)
2012-05-07 04:46:12 utc Mugatu so I would have assumed it's lifecycle ended almost immediately
2012-05-07 04:46:48 utc jmettraux I wonder how the issue can be simplified to its essence
2012-05-07 04:47:19 utc Mugatu I'll work up a small test case
2012-05-07 04:47:23 utc Mugatu be back in a bit :)
2012-05-07 04:47:29 utc jmettraux :-)
2012-05-07 05:39:03 utc Mugatu OK, this seems to demonstrate the issue: https://gist.github.com/9a11fd3cd561e096418e I took a while trying a number of variations :)
2012-05-07 05:40:29 utc jmettraux so the symptom is "trailing notifier expression" ?
2012-05-07 05:42:10 utc Mugatu actually, in this case, it seems to be the 'form_draft_reminder' expression that's hanging around
2012-05-07 05:42:19 utc Mugatu (which wraps the notifier)
2012-05-07 05:42:37 utc Mugatu if I'm reading the expression def correctly :)
2012-05-07 05:43:41 utc Mugatu but this is similar to what I saw before in the 'real' app, in that this expression is in a state of 'cancelling'
2012-05-07 05:46:12 utc jmettraux ok, ruote master?
2012-05-07 05:47:05 utc jmettraux https://github.com/jmettraux/ruote/issues/47
2012-05-07 05:48:26 utc Mugatu sorry, didn't understand the question for a minute -- yes, I'm working from ruote master :)
2012-05-07 05:50:19 utc jmettraux I'll take some time to investigate that later in the afternoon, meanwhile did you have a chance to test the small tentative fix I did for ruote-fluo?
2012-05-07 05:51:18 utc Mugatu ah no, I hadn't noticed that yet -- I'll give it a go
2012-05-07 06:03:03 utc Mugatu jmettraux: That fix works great in Chrome 19
2012-05-07 06:03:17 utc jmettraux Excellent, many thanks!
2012-05-07 06:03:43 utc Mugatu oh no, thank you! :)
2012-05-07 06:05:21 utc Mugatu OK, it's 2am here, I think I have to give up :)
2012-05-07 06:05:25 utc Mugatu thanks for your help as usual, John
2012-05-07 06:05:52 utc jmettraux you're welcome, have a good night!