<JordanBrown>
Like I said in the comments, I'm not sure that that binary-op PR is a good idea, but it's simple enough and I'm tired of listening to people say they want it.
MatrixBot1234514 has quit [Server closed connection]
MatrixBot1234514 has joined #openscad
cart_ has joined #openscad
rawgreaze has quit [Ping timeout: 268 seconds]
rawgreaze has joined #openscad
<teepee>
JordanBrown_: what happens with non-integer values?
noonien8 has quit [Server closed connection]
noonien8 has joined #openscad
<teepee>
hex constants are a breaking change, but I'm mostly ok with doing that, maybe adding a deprecated warning for names starting with number
greenbigfrog has quit [Ping timeout: 240 seconds]
cart_ has quit [Quit: Konversation terminated!]
greenbigfrog has joined #openscad
mmu_man has quit [Ping timeout: 246 seconds]
hyperair has joined #openscad
mmu_man has joined #openscad
<InPhase>
teepee: 0x03 = 3.14159265358979; What could go wrong.
<teepee>
indeed
califax has quit [Remote host closed the connection]
califax has joined #openscad
L29Ah has joined #openscad
<teepee>
we are sure someone already used that in production code, right?
califax has quit [Remote host closed the connection]
califax has joined #openscad
J23k11 has quit [Quit: Client closed]
J23k11 has joined #openscad
mmu_man has quit [Ping timeout: 256 seconds]
snaked has quit [Quit: Leaving]
cart_ has joined #openscad
mmu_man has joined #openscad
J23k11 has quit [Quit: Client closed]
J23k11 has joined #openscad
mmu_man has quit [Ping timeout: 255 seconds]
h-erectus has joined #openscad
erectus has quit [Remote host closed the connection]
<Scopeuk>
I would be more surprised if no one did 0x10 =16; than if someone has done it to "get hex constants"
<crole>
dalias: I really don't see anyone saying F360 is "good at cad." That's the issue to me.
<InPhase>
crole: There are slightly more people saying "Fusion 360 is good" than "OpenSCAD is good", but if you take that as a proportion of the total times Fusion 360 or OpenSCAD are mentioned anywhere on the Internet, the positive statements about OpenSCAD seem to be coming out way ahead.
<InPhase>
So we can deduce a significantly larger Fusion 360 userbase, but that OpenSCAD seems to be far more appreciated by its users.
<InPhase>
These results do not shock me. :)
GNUmoon has quit [Remote host closed the connection]
GNUmoon has joined #openscad
guso78k has joined #openscad
<crole>
That's how I feel, and I'll kill anyone who tries to say "well, once you get past all of the extremely anti-user behaviors AutoDesk has..."
misterfish has quit [Ping timeout: 268 seconds]
misterfish has joined #openscad
mmu_man has joined #openscad
guso78k has quit [Ping timeout: 250 seconds]
Lagopus has joined #openscad
teepee_ has joined #openscad
teepee has quit [Ping timeout: 264 seconds]
teepee_ is now known as teepee
<JordanBrown_>
teepee: internal detail, the doubles are cast into uint32_t, so C conversion semantics apply. At an external level, fractions are dropped. I don't know offhand (and am sitting at the wrong computer to test easily) whether negative non-integers truncate toward zero or toward negative infinity.
<JordanBrown_>
WRT it being a breaking change... I hadn't realized that you could have variable names that started with a digit. That's just gross. But you're right, it is. We could make that breaking change, or we could have a hex-string-to-number function.
<JordanBrown_>
Going back to conversion, without looking at the standards I'm not 100% sure that C fully specifies the double-to-integer conversion, so in theory if it doesn't behavior might differ between platforms. Don't do that. Or we could push the double through floor() before converting.
<dalias>
it does except when it's undefined
<dalias>
(undefined on overflow, nans, etc.)
<dalias>
but if the value is in-range, it's 100% specified (truncation)
<JordanBrown_>
I dimly remember that there's some rounding behavior that's platform-defined.
<dalias>
not in conversion to integer
<JordanBrown_>
I'll go look.
<dalias>
if it's not conforming to annex f (not ieee arithmetic), floating point rounding could behave in platform-specific ways
<dalias>
but for annex f conforming implementations, even that is 100% defined
<dalias>
(modulo the rules about excess precision -- all floating point expressions take place in a type that might have more precision than the nominal type of the expression. you can get direct access to this using double_t and float_t instead of double and float. but in practice it only applies on 32-bit x86 and m68k. nothing else has excess precision.)
<peeps[zen]>
JordanBrown_: it would be good to add the ops to some tests or possibly create a new one. there are some that do a number of tests across an array of values, such as tests/data/scad/misc/expression-evaluation-tests.scad
<peeps[zen]>
probably some other similar ones under misc there too
<teepee>
peeps[zen]: no crash with gcc-13 but compile warning treated as error, that should be fixable via extra compile option
<peeps[zen]>
teepee: ah, did you build the same intel-tbb version from mxe still or are you using pca's patch, or something else?
<JordanBrown_>
It looks like conversions from 64-bit int to double round in a platform-defined way when the value is greater than can be precisely represented, so that's a theoretical concern if we ever go greater than 32-bit.
<JordanBrown_>
peeps[zen] good point about tests.
<JordanBrown_>
The actual arithmetic is pretty simple, so I'm not worried about it, but tests would give warm fuzzy feelings that we do in fact get consistent results across platforms.
<peeps[zen]>
teepee: so you made a new mxe package for onetbb? how are you dealing with linking?
<teepee>
nothing yet, so far the build did not even return, not sure what it is doing so long
<JordanBrown_>
real work now :-)
<JordanBrown_>
later...
<peeps[zen]>
teepee: that seemed like a big issue to me, that the CMakeLists warns that its not supported if you try to build static and may cause problems.
Guest18 has joined #openscad
paddymahoney has quit [Server closed connection]
<peeps[zen]>
i guess we could technically have mixed linking of shared/static for different packages? but the scripts are currently set up for all or nothing, and i think it would complicate packaging a bit
paddymahoney has joined #openscad
Guest18 has quit [Client Quit]
Guest41 has joined #openscad
Guest41 has quit [Client Quit]
<JordanBrown_>
How do people feel about the current behavior where "opening" some file types (e.g. STL, PNG) inserts import() or other references to the file? I'm not sure how you trigger that behavior, since the normal open dialog restricts file types to scad and csg, but there's implementation. The reason that
<JordanBrown_>
for low-hanging fruit and started to work on the "let me edit random file types in your editor" one. Easy enough, but there's a conflict on handling of those magic types when they are text formats like ASCII STL and SVG.
<JordanBrown_>
I asked is that I glanced through issues looking
teepee has quit [Ping timeout: 264 seconds]
teepee has joined #openscad
L29Ah has quit [Ping timeout: 255 seconds]
<peeps[zen]>
JordanBrown_: trigger by opening gui from command line i suppose
L29Ah has joined #openscad
<peeps[zen]>
imo it couldn't hurt to allow a "All File Types .*" secondary selection in the GUI open dialog
<peeps[zen]>
JordanBrown_: btw just realized why your message was so hard to parse the 2nd and 3rd part are swapped on my end
<peeps[zen]>
i would probably just prioritize magic type check before checking for text format. if someone wants to edit ascii stl or svg they can fall back to a regular text editor
<peeps[zen]>
having auto import seems like a decent convenience for previewing an stl etc. not that I've ever really used it
cart_ has quit [Ping timeout: 276 seconds]
mmu_man has joined #openscad
misterfish has quit [Ping timeout: 255 seconds]
J23k11 has quit [Quit: Client closed]
J23k11 has joined #openscad
<JordanBrown_>
BTW, I had really puzzling results when I simply added "All file types" to the open dialog. Sometimes it would work and sometimes not. I isolated it to this "known file types" mechanism and it was still puzzling until I added diagnostics to dump the known-file-type list, and found that despite the
<JordanBrown_>
complete lack of obvious dynamic changes, it was
<JordanBrown_>
changing over time, adding file types. Turns out that if you ask QMap to look up an entry that isn't there, it creates one...
<JordanBrown_>
The existing logic would use .contains() to check to see if the type was known, but would also unconditionally attempt to retrieve the value, so the first time it was unknown but the second time it was known.
mmu_man has quit [Ping timeout: 255 seconds]
JordanBrown has joined #openscad
JordanBrown has quit [Ping timeout: 255 seconds]
mmu_man has joined #openscad
misterfish has joined #openscad
hisacro has quit [Server closed connection]
hisacro has joined #openscad
GeneR has joined #openscad
<GeneR>
I have a question. Have people seen the manifold engine creating non-manifold objects when doing boolean operations with manifold objects?
GeneR has quit [Quit: Client closed]
GeneR has joined #openscad
<GeneR>
I have a question. Have people seen the manifold engine creating non-manifold objects when doing boolean operations with manifold objects?
<GeneR>
I used the manifold engine because the CSG engine takes tens of minutes on an i7 MacBook Pro
teepee has quit [Remote host closed the connection]
<GeneR>
My program has now been running for about an hour of CPU time. Is there a way to submit programs or some way I can see where in the process it is hung up?
teepee has joined #openscad
<JordanBrown_>
Here's a terrible idea for replacing 0xHHH syntax in that binary-ops PR: have a function hex(). When you give it a string, it converts it to a number... and when you give it a number it converts it to a string.
linext has joined #openscad
berndj has quit [Server closed connection]
berndj has joined #openscad
GeneR has quit [Quit: Client closed]
teepee has quit [*.net *.split]
GNUmoon has quit [*.net *.split]
h-erectus has quit [*.net *.split]
califax has quit [*.net *.split]
aiyion has quit [*.net *.split]
TheAssassin has quit [*.net *.split]
GNUmoon has joined #openscad
erectus has joined #openscad
califax has joined #openscad
<JordanBrown_>
GeneR: depends a bit on the circumstances. If it's an F6 render and it's stuck in the actual rendering, there's a progress bar and nnn/1000 counter. (That's not at all linear; the final steps usually take the most time.)
aiyion has joined #openscad
TheAssassin has joined #openscad
teepee has joined #openscad
<JordanBrown_>
If it's the actual program execution, there's nothing other than whatever you might echo().
<JordanBrown_>
Generally, I don't think F5 preview gets progress reporting.