whitequark changed the topic of #solvespace to: SolveSpace--parametric 2d/3d CAD · latest version 2.2 · http://solvespace.com · code at https://github.com/solvespace/solvespace · logs at https://irclog.whitequark.org/solvespace
mumptai has quit [Ping timeout: 250 seconds]
mumptai has joined #solvespace
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined #solvespace
mumptai_ has joined #solvespace
<GitHub> [solvespace] michthom opened issue #96: Mac OS crash (EXC_BAD_ACCESS) https://github.com/solvespace/solvespace/issues/96
<GitHub> [solvespace] michthom commented on issue #96: Version 3.0 (3.0~ea52fcbc) compiled from source with debug symbols. https://github.com/solvespace/solvespace/issues/96#issuecomment-258600782
<GitHub> [solvespace] whitequark commented on issue #96: Please run SolveSpace under debugger and tell me the values of locals `r` and `c` at the moment of crash. https://github.com/solvespace/solvespace/issues/96#issuecomment-258600858
<GitHub> [solvespace] michthom commented on issue #96: This is the epitome of speedy service! I was running it under LLDB, so when it stopped I dumped the "thread backtrace all" above. Stupidly I didn't then write a core file to analyse. I'll try to reproduce it now. https://github.com/solvespace/solvespace/issues/96#issuecomment-258600971
mumptai_ has quit [Quit: Verlassend]
_whitelogger has joined #solvespace
whitequark has joined #solvespace
<ohsix> so was reading some of the bugs and saw there's some type of expression support, reading the code and mashing buttons in constraints
<ohsix> so far pi and pi+/* / numbers work, but i see stuff for names, where can i enter those
<whitequark> you can't
<whitequark> there's only i
<whitequark> pi
<ohsix> oh i see
<ohsix> i saw the name parsing code and didn't register that pi/sin/whatever followed
<ohsix> and underscore is just in there in case some_values or expressions show up later?
<ohsix> longjmp is going to leak across all those edges that call AllocExpr :p
<whitequark> nope
<whitequark> AllocExpr uses an arena allocator
<whitequark> see FreeAllTemporary
<ohsix> nice
<ohsix> yea
<whitequark> we in fact run solvespace under valgrind and other memory leak checkers
<ohsix> it was kind of a joke
<whitequark> ah
<ohsix> i'm math dumb, where's the actual solver-iest bit of the solver located
<ohsix> and is there a name or citation i can start reading to see why i couldn't do the obvious to me thing, and just put the expressions in place of the value and have it work
<ohsix> ignoring loops and other dependencies of course, just trivial cases
<whitequark> constrainteq.cpp, system.cpp
<whitequark> expressions in place of value?
<ohsix> yea, you evaluate an expression on entry, 'merely' letting you reference a value and keeping Expr around for the constraint, evaluating it in the solver
<ohsix> i'm sure there's an obvious way it won't work but i don't have a great math background
<ohsix> way/reason
<whitequark> no, there's no real reason what you say can't be implemented
<whitequark> it just... isnt
<whitequark> mostly the UI is missing
<ohsix> oh, nice
<ohsix> k, i misread some stuff that implied numerical problems
<ohsix> not a hueg fan of c++ but this looks like a decent application of it
<whitequark> c++ is atrocious
<ohsix> like for vectors and quaternions and basic things
<ohsix> yea
<whitequark> we've had to fight it constantly
<whitequark> anyway
<whitequark> a full rewrite would not happen with solvespace, so we're stuck with c++
<ohsix> there usually aren't a lot of is-a relationships that make it broad enough to be worth doing, and people try and force it with weird objects all over and extensibility that isn't strictly called for
<whitequark> no point in discussing it much
<whitequark> yeah
<whitequark> one of the many issues
<ohsix> oh, no argument from me, it was a compliment
<ohsix> it was one of the few chances i actually get to say it huhuhu
<ohsix> it was fun to model a laser stage and a pantograph by just doing the obvious
<ohsix> aside from the line i had to put in the drawing just to keep the arcs equal length ;]
<ohsix> there's two operations you can do when you constrain length/distance/radius for an arc, and i wanted length, it took a bit to figure out
<whitequark> yeah
<whitequark> screensoht?
<ohsix> perhaps i was talking it up a bit .
<ohsix> heheh but the motor is the actuator that moves the mirror, the angle of the laser just changes with respect to the 'mirror', i did it as an illustration for a friend
<ohsix> i couldn't think of a way to actuate it on a pivot in plane, center with the mirror like a galvanometer would
<ohsix> that's why i did the pantograph, it's got a rotate-y thing
<ohsix> it has a long sliding handle and doesn't go in a full revolution :]
<whitequark> oh i see
<whitequark> >maxim chip datasheet
<ohsix> i think i mentioned this last time, but i can build a paper model and it might help teaching
<whitequark> you are so innocent :D
<ohsix> yes that's to remind me i am putting something off
<ohsix> i don't really need the datasheet anymore, figured that part otu a while ago and started messing with the fx2 logic analyzer thing instead
<ohsix> been doign electronics since i was a kid but only recently has ebay and china made it really cheap for me to kit up on stuff
<ohsix> this isn't really a request or problem, but the arc thing, since it is ambiguous which you might want, could/should there be ui asking you which? or otherwise letting you choose
<ohsix> concievably you'd want both
<GitHub> [solvespace] michthom commented on issue #96: Reproduced below, and saved a core file this time - but I'm clueless on how to display the locals you wanted, I tried (to no avail):... https://github.com/solvespace/solvespace/issues/96#issuecomment-258607304
<GitHub> [solvespace] whitequark commented on issue #96: Reproduced below, and saved a core file this time - but I'm clueless on how to display the locals you wanted, I tried (to no avail):... https://github.com/solvespace/solvespace/issues/96#issuecomment-258607304
<GitHub> [solvespace] whitequark commented on issue #96: You need to go up one frame before asking lldb for this variable. https://github.com/solvespace/solvespace/issues/96#issuecomment-258610061
<GitHub> [solvespace] michthom commented on issue #96: Got it, sorry for being slow:... https://github.com/solvespace/solvespace/issues/96#issuecomment-258617583
<GitHub> [solvespace] michthom commented on issue #96: Got it, sorry for being slow:... https://github.com/solvespace/solvespace/issues/96#issuecomment-258617583
<GitHub> [solvespace] michthom commented on issue #96: Got it ( at last! )... https://github.com/solvespace/solvespace/issues/96#issuecomment-258617942
<GitHub> [solvespace] whitequark commented on issue #96: Got it ( at last! )... https://github.com/solvespace/solvespace/issues/96#issuecomment-258617942
<GitHub> [solvespace] whitequark commented on issue #96: Ah yes... `c = -128`. https://github.com/solvespace/solvespace/issues/96#issuecomment-258624720
<GitHub> [solvespace] whitequark pushed 1 new commit to master: https://github.com/solvespace/solvespace/commit/afecbccb0f89e9288c20a06ace8bea998ec3aac7
<GitHub> solvespace/master afecbcc whitequark: OS X: fix another out-of-bounds TW.META access....
<travis-ci> solvespace/solvespace#167 (master - afecbcc : whitequark): The build passed.
<GitHub> [solvespace] whitequark commented on issue #96: Fixed in master and 2.x. https://github.com/solvespace/solvespace/issues/96#issuecomment-258627286