mitevpi / thesaurus

TT Hackathon 2018 - Autocomplete for Visual Programming Nodes
MIT License
25 stars 7 forks source link

Need Error-Catching on First Run #9

Open johnpierson opened 5 years ago

johnpierson commented 5 years ago

If you install thesarus and attempt to click show/hide without first training the data. You get a nasty error. image

Could be good to have a simple message that says something like,

Hi there! Before showing thesaurus we need to train some data.

I am presenting on this extension at BILTANZ and will mention that consideration though! (Will also see if I can actually get around to contributing to this repo)

ksobon commented 5 years ago

@johnpierson thanks for sharing mate. I agree we should do a quick path validation to make sure that the training file exists before we enable that UI.

QilongTang commented 5 years ago

@johnpierson Thank you for raising this. Definitely agree there. Also would love your feedback on how useful it is right now for people, I keep thinking maybe people should not know about the training level at all because all the other ML tool I have seen they do not ask for people to train module themselves.

ksobon commented 5 years ago

Well, that's because we wanted to allow people to train the module on "their" dynamo graphs so that the model is geared towards how they work, and not how me or you work.

johnpierson commented 5 years ago

Also would love your feedback on how useful it is right now for people, I should hopefully have some after BILTANZ! (May 23-25).

I agree with @ksobon , I don't think it is necessary to remove that for people. I like the idea of having them train the data based on their needs. :)

We simply need a bit more transparency on the steps I think

QilongTang commented 5 years ago

@johnpierson @ksobon Reason I am asking because download number wise I did not it has been fully utilized at the level I imagined, so I was wonder if it is because it is currently too hard to pick up as a new tool for them.

johnpierson commented 5 years ago

@QilongTang I actually think it may be due to slow Dynamo 2.0 adoption. With 2.1 being the standard in 2020, I think you all are going to see a lot more adoption.

Like I mentioned, I am showing the tool at BILTANZ on the 24th. So it should hopefully drum up some additional interest.

ksobon commented 5 years ago

Reason I am asking because download number wise I did not it has been fully utilized at the level I imagined, so I was wonder if it is because it is currently too hard to pick up as a new tool for them.

I would imagine that we can do a better job marketing the tool a little and I am sure people would pick it up...mostly because the current search menu is such a crappy option. Sorry to say that, but god knows no one uses that. Also, @QilongTang your proposal for docking panels is a good one. That would for sure help adoption.