teepee changed the topic of #openscad to: OpenSCAD - The Programmers Solid 3D CAD Modeller | This channel is logged! | Website: http://www.openscad.org/ | FAQ: https://goo.gl/pcT7y3 | Request features / report bugs: https://goo.gl/lj0JRI | Tutorial: https://bit.ly/37P6z0B | Books: https://bit.ly/3xlLcQq | FOSDEM 2020: https://bit.ly/35xZGy6 | Logs: https://bit.ly/32MfbH5
noonien808310429 has quit [Ping timeout: 268 seconds]
kintel has joined #openscad
<gbruno> [github] kintel closed pull request #5071 (Bugfix: Color schemes no loading due to uninitialized pointer) https://github.com/openscad/openscad/pull/5071
<gbruno> [github] kintel pushed 1 modifications (Bugfix: Color schemes no loading due to uninitialized pointer (#5071)) https://github.com/openscad/openscad/commit/f2d4c1b62843e2d90e366b690c68849b4101df0c
<gbruno> [github] kintel closed issue #5070 (Does not see user color schemes (Mac OS 13.6.4)) https://github.com/openscad/openscad/issues/5070
<kintel> pca006132 Yeah, I believe it's a straight-up LRU cache
mmu_man has quit [Ping timeout: 268 seconds]
mmu_man has joined #openscad
drfff has joined #openscad
guerd87 has quit [Remote host closed the connection]
guerd87 has joined #openscad
LordOfBikes has quit [Ping timeout: 256 seconds]
LordOfBikes has joined #openscad
KirubelG80 has joined #openscad
KirubelG80 has quit [Client Quit]
JordanBrown has quit [Read error: Connection reset by peer]
JordanBrown1 has joined #openscad
LordOfBikes has quit [Ping timeout: 268 seconds]
t4nk_fn has quit [Remote host closed the connection]
t4nk_freenode has joined #openscad
t4nk_freenode is now known as t4nk_fn
LordOfBikes has joined #openscad
mmu_man has quit [Ping timeout: 268 seconds]
J24k15 has joined #openscad
J24k25 has quit [Ping timeout: 250 seconds]
<pca006132> kintel: so essentially it is throwing stuff away until the cache size is under control?
<kintel> yup
<kintel> Not the most elegant solution, but it did a reasonable job for CGAL.
<kintel> ..and we stole the cache class from Qt ;)
<pca006132> I think with manifold, it makes sense to only cache things like minkowski
<pca006132> or crazy unions/intersection, idk
<pca006132> feel like it is not worth it to cache anything else
<pca006132> especially since caching may prohibit manifold from doing some optimization
<pca006132> but avoid caching intermediate nodes may be enough, will probably require some *interactive* benchmarks
<kintel> In terms of "prohibit manifold from doing some optimization":
<kintel> Are you saying that just holding a reference to a Manifold object may prohibit such optimization?
<pca006132> yes
<pca006132> well, it is quite complicated, maybe you should look at the code, let me find a reference to that
<pca006132> the reason is that if we flatten the children when there are other references, we may end up evaluating the same thing twice
<kintel> This all happens at the time of resolving the root node of the tree, right?
<pca006132> yes
<pca006132> we may be able to do something better, but it is hard to make it fast under all cases
<pca006132> another interesting thing for openscad: if openscad can do something like common subexpression elimination, so the csg 'tree' becomes a DAG, we can avoid evaluating the shared nodes multiple times
<kintel> The cache is kind of the mechanism we use for that, but ideally we'd do it as a separate optimization pass instead
<kintel> (the cache key is the normalized string representation of the OpenSCAD subtree after evaluating all variables and instantiating all modules)
<pca006132> I see
<kintel> Anyway, I understand the challenge. We should probably prefer caching concrete geometry rather than Manifold trees (PolySet, MeshGL etc.)
<pca006132> yup
<pca006132> caching the unevaluated promise is not very useful :P
<kintel> ok, needs more work. We still need a measure of cost and size.
<pca006132> I can probably open a PR this weekend
<pca006132> I am still going through the horrible midterm exam and assignments from the students
<pca006132> I feel that the students are trying to murder me with their submissions...
<kintel> We can estimate cost by measuring real time, and if we cache concrete geometry we have a concrete size.
<pca006132> do note that the time can be multithreaded though
<gbruno> [github] kintel edited issue #4825 (Move manifold out of experimental) https://github.com/openscad/openscad/issues/4825
<kintel> yeah, but that's fine. Fast system with many cores -> less need for caching :)
<pca006132> true
<InPhase> pca006132: That was my least favorite part of that job. :)
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pca006132> we thought that asking them to submit LaTeX document will make sure that the result is readable
<InPhase> lol
<pca006132> and then some of them submitted things that are all bold, or all monospace, or zero page margin
<pca006132> me: ???
<InPhase> Now you know to provide a template for next year...
<pca006132> no, for the next assignment :)
<InPhase> Or that. :)
<pca006132> but I still have no idea why some of them will submit something like that
<pca006132> it takes effort to make things all bold or monospace or remove the page margin
<pca006132> do they think it is more legible?
<pca006132> or they just try to harm me :(
<InPhase> Did they submit the latex source as well?
<pca006132> no
<InPhase> Maybe they produced their pdfs in some other horrible way, like writing it in a text editor intending to make it LaTeX, and then printing as pdf...
<pca006132> oh god...
<pca006132> and some of them wrote text in math mode without \text
<pca006132> my sanity was decreasing when I was grading the assignments
<InPhase> It typically takes an entire pass through a course to fully measure the distribution of what is unknown.
<pca006132> yeah, I think we had too high expectation on them...
<pca006132> we thought that this is supposed to be an honor course, with course grade requirements on some prerequisite, so they should not be this bad
HumanG33k has quit [Ping timeout: 256 seconds]
HumanG33k has joined #openscad
Scopeuk has quit [Quit: Ping timeout (120 seconds)]
Scopeuk has joined #openscad
califax has quit [Remote host closed the connection]
califax has joined #openscad
califax has quit [Remote host closed the connection]
califax has joined #openscad
cart_ has joined #openscad
aiyion3 has quit [Remote host closed the connection]
erectus has quit [Remote host closed the connection]
aiyion3 has joined #openscad
erectus has joined #openscad
mmu_man has joined #openscad
snaked has quit [Quit: Leaving]
cart_ has quit [Ping timeout: 255 seconds]
gimzmoe has quit [Quit: WeeChat 3.5]
gimzmoe has joined #openscad
Guest30 has joined #openscad
<Guest30> hey can someone help me adjust a few prints i need the exterior dimensions enlarged and the interior cuts kept the same size
cart_ has joined #openscad
lostapathy has quit [Quit: WeeChat 3.5]
lostapathy has joined #openscad
<Guest30> hey can someone help me adjust a few prints i need the exterior dimensions enlarged and the interior cuts kept the same size
Guest30 has quit [Quit: Client closed]
pca006132 has quit [Remote host closed the connection]
R2robot has quit [Ping timeout: 268 seconds]
RoyK has quit [Ping timeout: 252 seconds]
RoyK has joined #openscad
<J24k15> Kintel i just rendered a huge array of cubes (20k)  memory was fine but it locked cpu at 90% for 10minutes  .. i remember we had a issue with a sphere and spikes that killed memory but this was pure cpu
<J24k15> ok nevermind fast csg takes longer
<J24k15> 9min 50 .. but with less CPU load
cart_ has quit [Ping timeout: 272 seconds]
<J24k15> so shouldn't manifold be faster using all cores not just one?
<J24k15> a quarter (about 5k cubes) renders in 20 sec
<J24k15> So i guess something happens that shouldn't or is exponential time expected (the last 1% probably is the final union?)
mmu_man has quit [Ping timeout: 255 seconds]
guso78k has joined #openscad
guso78k has quit [Quit: Client closed]
mmu_man has joined #openscad
R2robot has joined #openscad
L29Ah has quit [Ping timeout: 268 seconds]
L29Ah has joined #openscad
J24k15 has quit [Quit: Client closed]
J24k15 has joined #openscad
R2robot has quit [Quit: Pull the lever, Kronk. Wrong leverrrrrrr!]
R2robot has joined #openscad
L29Ah has quit [Excess Flood]
L29Ah has joined #openscad
<J24k15> Just got an email from github with a wrong name - regarding a login attempt (which was mine)  did someone else ever got something like that.
<J24k15> in the past all emails had my account name and now they writing  " Hi Jack"  that name was never used with github
<J24k15> it is like an AI scanned the internet and found a connection between my email and the nick "jack"
mmu_man has quit [Ping timeout: 268 seconds]
mmu_man has joined #openscad
<peepsalot> sounds like a phishing scheme
<InPhase> Most essential is to never enter information after clicking an emailed link.
<InPhase> Except in the unfortunately contradictory case of password reset emails, where you're counting on narrow time windows to protect you.
<J24k15> no link .. they send me the verification number
<InPhase> So do you ever go by the name Jack online?
<J24k15> and as i got this a few seconds after i opened github (in the wrong browser) i think it was really from github
<J24k15> that is the point not with github .. on printables  but they also added a part of a number to the name - which is part of my password which is very strange for a coincident
<J24k15> that is why i can only think of an AI having access to all these databases
<InPhase> If it's really github, then I guess search your github settings to see if things are set wrong in there? It sounds weird to me.
<InPhase> I don't get verification emails from github, as I switched it to totp by default.
<J24k15> i have searched all my profile data .. which is mostly blank and couldn't find anything
<J24k15> and all other emails from github where to the registered email and to my account name
<InPhase> J24k15: How about in a microsoft account?
<InPhase> Maybe some foolish executive decided to start cross-populating data...
<J24k15> i don't have any .. except the local account on my pc
<InPhase> Are you Jack there? :)
<J24k15> yes Ü .. but without the number
<J24k15> Ok seems this is solved ..  i have an empty 10years old never used github account
<InPhase> Ah. :)
teepee_ has joined #openscad
teepee has quit [Ping timeout: 260 seconds]
teepee_ is now known as teepee
Pb| has joined #openscad
L29Ah has quit [Read error: Connection reset by peer]
L29Ah has joined #openscad
mmu_man has quit [Ping timeout: 252 seconds]
mmu_man has joined #openscad
Pb| has quit [Quit: Client closed]
foul_owl has quit [Ping timeout: 255 seconds]
foul_owl has joined #openscad