利用者:Sebbas/GSoC 2016/Reports/Week 8

Week 8

What's been done

• So this week I bumped into a discussion on IRC on building the fluid-mantaflow branch on Windows. I had always mentally postponed the Windows build but apparently this week reality caught up.

• Building itself was fairly uncomplicated. There was just the old smoke library that had to be removed. (rB8eac981888df)

• The main issue under Windows was (and still is) that it is unable to run the Mantaflow script snippets. My first guess was that I would have to load Mantaflow scripts, the strings themselves, differently (rB4ffac9faf19c). But this did not help the cause.

• Eventually I figured that the problem with Windows is the static Mantaflow linking. Under Mac and Linux we can achieve this with the appropriate flags (i.e. --whole-archive). For Windows, however, I did not find a similar option and to my understanding one has to dynamically link a lib (i.e. using .dll).

Next up

• Just as last week the idea is to get the UI ready. I have already received some suggestions (thanks for that!) and will probably bring up one or the other topic in the #blenderui chat.

• Some university exams are coming up, so I'll try to balance that.

Questions

• I found the bmesh library setup quite interesting. It's split into python lib and internal lib. Maybe it would be a good idea to use such a setup for Mantaflow as well? I am not really happy with the static Mantaflow linking and maybe this can help the case.

Notes

• Windows users: It was nice looking into the Windows build but I think fixing it is not the top priority at the moment. For now, maybe you can try the Linux build?