It would be terribly late at night but I could make around 20 minutes. I would prefer it to be in around 45 minutes though.ickby wrote: I think it would be good to discuss a few technical things first, so that you get at least a first impression on the topic and what it is about. Are you on IRC/Gitter in 2hours? We could talk a bit about the technical side of things.
Proposal for GSoC - jnxd
Forum rules
Be nice to others! Respect the FreeCAD code of conduct!
Be nice to others! Respect the FreeCAD code of conduct!
Re: Proposal for GSoC - jnxd
Re: Proposal for GSoC - jnxd
Hello jnxd,
thanks for the small talk yesterday, better a short one than nothing I don't have time till monday for large sessions, but can definitely review the proposals you create. So I propose you do focus on the following things:
- Learning time. Imho learning about the problem itself and reading a few papers of solutions others have found to topologicalnaming shoul dbe included in your proposal
- Understanding the existent Code. I'm not saying you must use it as basis, if you find something more clever it is also ok. But understanding what is done is important, as only than a good comparison is possible.
- Setting up reference problems. Creating a set of small or large problems showing the topological naming issue to later show your algorithms effectifness is important. They should include small ones to be able to test your work wothout the need of a full stack implementation but to test each implementation step
- Coding, either extending the existing things or doing it new. You should provide clear insights in what you think you can achieve. You wont fix the whole problem in 3 months, but please write down to which state you like to proceed during FGSoC.
Stefan
thanks for the small talk yesterday, better a short one than nothing I don't have time till monday for large sessions, but can definitely review the proposals you create. So I propose you do focus on the following things:
- Learning time. Imho learning about the problem itself and reading a few papers of solutions others have found to topologicalnaming shoul dbe included in your proposal
- Understanding the existent Code. I'm not saying you must use it as basis, if you find something more clever it is also ok. But understanding what is done is important, as only than a good comparison is possible.
- Setting up reference problems. Creating a set of small or large problems showing the topological naming issue to later show your algorithms effectifness is important. They should include small ones to be able to test your work wothout the need of a full stack implementation but to test each implementation step
- Coding, either extending the existing things or doing it new. You should provide clear insights in what you think you can achieve. You wont fix the whole problem in 3 months, but please write down to which state you like to proceed during FGSoC.
Stefan
- DeepSOIC
- Veteran
- Posts: 7896
- Joined: Fri Aug 29, 2014 12:45 am
- Location: used to be Saint-Petersburg, Russia
Re: Proposal for GSoC - jnxd
Hi!
I'm in doubt if I want to mentor you or not. Your choice of task is unexpectedly ambitious, and we (me and ickby) think that the odds of failure are just too high.
So I'd like to ask a few questions.
1. Have you ever created a feature based on OCC shape from scratch? (either in C++ or in Python).
2. If you were to make a replica of Part Fuse in python, can you do it?
3. In 2, can you make it support colors, like c++ version does?
4. Do you have any preliminary idea of how should toponaming work and feel? In particulat, let's say I'm writing a new feature, say Part Connect, and I want it to support toponaming, what do I need to do, approximately? (Ignoring for a while what ickby and ezzieguywuf have already done. This plan doesn't have to be complete, or otherwise be any good.. just the sole fact of having an evil plan matters much already)
PS. Even if you don't get accepted, your attempt at toponaming will be highly appreciated. You can expect no less (and likely more )mentorship/assistance/suggestions/opinions if you do it outside of GSoC.
I'm in doubt if I want to mentor you or not. Your choice of task is unexpectedly ambitious, and we (me and ickby) think that the odds of failure are just too high.
So I'd like to ask a few questions.
1. Have you ever created a feature based on OCC shape from scratch? (either in C++ or in Python).
2. If you were to make a replica of Part Fuse in python, can you do it?
3. In 2, can you make it support colors, like c++ version does?
4. Do you have any preliminary idea of how should toponaming work and feel? In particulat, let's say I'm writing a new feature, say Part Connect, and I want it to support toponaming, what do I need to do, approximately? (Ignoring for a while what ickby and ezzieguywuf have already done. This plan doesn't have to be complete, or otherwise be any good.. just the sole fact of having an evil plan matters much already)
PS. Even if you don't get accepted, your attempt at toponaming will be highly appreciated. You can expect no less (and likely more )mentorship/assistance/suggestions/opinions if you do it outside of GSoC.
Re: Proposal for GSoC - jnxd
No, but I plan to in the meantime. I could follow the bottle tutorial. My efforts until recently were just constrained to adding the constraint modes in sketcher (I'm afraid to bring that up, given you chastized it so much ). I had moved to partdesign, but I doubt there much OCC involved yet.DeepSOIC wrote:Hi!
I'm in doubt if I want to mentor you or not. Your choice of task is unexpectedly ambitious, and we (me and ickby) think that the odds of failure are just too high.
So I'd like to ask a few questions.
1. Have you ever created a feature based on OCC shape from scratch? (either in C++ or in Python).
I would probably need a template. Do you mean a reimplimentation of the feature? How about we try doing that over this week and the next?2. If you were to make a replica of Part Fuse in python, can you do it?
3. In 2, can you make it support colors, like c++ version does?
I did get an idea from the paper by Agbodan, but I'm afraid I don't have anything solid right now.4. Do you have any preliminary idea of how should toponaming work and feel? In particulat, let's say I'm writing a new feature, say Part Connect, and I want it to support toponaming, what do I need to do, approximately? (Ignoring for a while what ickby and ezzieguywuf have already done. This plan doesn't have to be complete, or otherwise be any good.. just the sole fact of having an evil plan matters much already)
It is pretty much unlikely my proposal will be accepted, given the state that it is in at this time. But I do expect working on toponaming should help me in the long run, so I look forward to spend at least some time on this.PS. Even if you don't get accepted, your attempt at toponaming will be highly appreciated. You can expect no less (and likely more )mentorship/assistance/suggestions/opinions if you do it outside of GSoC.
- DeepSOIC
- Veteran
- Posts: 7896
- Joined: Fri Aug 29, 2014 12:45 am
- Location: used to be Saint-Petersburg, Russia
Re: Proposal for GSoC - jnxd
I know, I browsed your commits trying to get an impression of how well you know freecad's internals. The impression wasn't great, but impression can be wrong. So I decided to ask the questions.jnxd wrote:My efforts until recently were just constrained to adding the constraint modes in sketcher (I'm afraid to bring that up, given you chastized it so much ). I had moved to partdesign, but I doubt there much OCC involved yet.
As a user, I have no complaints about what you have done constraints now. You fixed what made me uncomfortable, I'm totally happy! In fact, if you applied for sketcher, or partdesign, or part, or maybe even a new workbench, I think you look no worse than Kurt!
I didn't look there, I have no idea what is the idea therejnxd wrote:I did get an idea from the paper by Agbodan,
Yes. Imagine you are to delete the class "Part::MultiFusion" (or however it is called...) and replace it with a python implementation.jnxd wrote:Do you mean a reimplimentation of the feature?
Re: Proposal for GSoC - jnxd
By python do you mean pythonocc? I haven't used it, but was just wondering if there is any other way of using occ with python.DeepSOIC wrote:s.
1. Have you ever created a feature based on OCC shape from scratch? (either in C++ or in Python).
- DeepSOIC
- Veteran
- Posts: 7896
- Joined: Fri Aug 29, 2014 12:45 am
- Location: used to be Saint-Petersburg, Russia
Re: Proposal for GSoC - jnxd
No, just using FreeCAD python API.
- DeepSOIC
- Veteran
- Posts: 7896
- Joined: Fri Aug 29, 2014 12:45 am
- Location: used to be Saint-Petersburg, Russia
Re: Proposal for GSoC - jnxd
Actually, I forgot, that you already answered:
Some amount of learning required for a task is very good, IMO. But in this case, I think the barrier of learning is too high now (freecad, OCC's toponaming, unfinished code that even I am afraid of digging). So judging "If I were you", I would certainly answer that it's very unlikely I will get anywhere any time soon. The problem will likely be a burden, and might push you away from FreeCAD instead of sucking you in.
I wonder if one can do actual code outside of the proposal, and still be in gsoc...? Ickby? Anyone?
So of course, you will eventually come up with the answer to second one, but I already understand that you don't know the picture of where you are supposed to introduce toponaming.jnxd wrote:1. Have you ever created a feature based on OCC shape from scratch? (either in C++ or in Python).
--No, but I plan to in the meantime.
Some amount of learning required for a task is very good, IMO. But in this case, I think the barrier of learning is too high now (freecad, OCC's toponaming, unfinished code that even I am afraid of digging). So judging "If I were you", I would certainly answer that it's very unlikely I will get anywhere any time soon. The problem will likely be a burden, and might push you away from FreeCAD instead of sucking you in.
I wonder if one can do actual code outside of the proposal, and still be in gsoc...? Ickby? Anyone?
- DeepSOIC
- Veteran
- Posts: 7896
- Joined: Fri Aug 29, 2014 12:45 am
- Location: used to be Saint-Petersburg, Russia
Re: Proposal for GSoC - jnxd
Actually, I have an idea. How about having a start by adding an interesting feature to Part/PartDesign, that is just a little bit related to toponaming? I think it will be more-or-less within the proposal, yet fun! I am pressing that "want to mentor" button
Re: Proposal for GSoC - jnxd
Glad to see you on board! However, what feature would you suggest to add related to TN?DeepSOIC wrote:Actually, I have an idea. How about having a start by adding an interesting feature to Part/PartDesign, that is just a little bit related to toponaming? I think it will be more-or-less within the proposal, yet fun! I am pressing that "want to mentor" button