<JordanBrown>
Just built the current sources. When I start the program and click "New", it seems to do a preview of the empty file.
<JordanBrown>
And no, Design/Automatic Reload and Preview is not set.
<JordanBrown>
That's "New" in the launcher window.
fling has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
fling has joined #openscad
<JordanBrown>
But if the Welcome Screen (launcher window) is disabled, it doesn't.
mmu_man has quit [Ping timeout: 276 seconds]
<JordanBrown>
Even more strangely, sometimes it does and sometimes it doesn't.
mmu_man has joined #openscad
sparsh-n has joined #openscad
<sparsh-n>
Hi all, this is a bit late, but looks like the Windows Build/CI project is in for GSoC! I look forward to working on OpenSCAD this summer! As for the quaternion rotation uses I mentioned earlier this month, I think the primary benefit was that it avoid gimbal locking, and are more compact in terms of representation.
sparsh-n has quit [Quit: Client closed]
<InPhase>
JordanBrown: I'm having trouble replicating that one. But the related longstanding minor nuissance I've observed is that whenever I open an existing file, it will preview on the first character entered in the editor window. Which unless that is a whitespace character, always means a syntax error.
mmu_man has quit [Ping timeout: 276 seconds]
<InPhase>
JordanBrown: I don't use the editor regularly, so I don't actually get clobbered by this very often. But it's definitely a buggy behavior.
<InPhase>
On a slow previewing model, this would surely be annoying to anyone getting hit by it repeatedly because they use the editor.
<InPhase>
Simply pressing the space bar could result in a long delay.
<InPhase>
Ironically any other non-whitespace character would not be a long delay, because that would cause a syntax error and terminate quickly... But users should not have to do a weird dance. lol
<InPhase>
Not sure if that one ever got marked as an issue. I think it's been around for a while though, because I just tested and see the same behavior in the last release.
<JordanBrown>
That behavior appears associated with Design/Automatic Reload and Preview.
<JordanBrown>
Can you confirm that you only see it when that's set?
<InPhase>
Yes, it seems definitely associated with that.
<InPhase>
Perhaps it has gone unreported since most people with that set are not using the editor regularly anyway.
<InPhase>
But it becomes wonky if someone switches over then.
<JordanBrown>
From the outside, it looks to me like it's triggering a preview any time the buffer changes and the file was previously saved or opened.
<InPhase>
Yeah, seems to be the case.
<JordanBrown>
That is, it does *not* trigger if you have a new never-saved file.
<InPhase>
Right.
<InPhase>
Perhaps on the first buffer change it is switching files to the temporary one, triggering a new reload and preview?
<InPhase>
(Which would need suppressed somehow.)
<JordanBrown>
By "the temporary one" you mean the backup file?
<InPhase>
Yes.
<InPhase>
I thought that was switched at the trigger of preview, but maybe it's done at the first buffer edit.
<InPhase>
s/switched/copied/
<JordanBrown>
If I have a new (never saved, never opened) file then it does not trigger this behavior even when you've previewed it and so saved to a backup file.
<JordanBrown>
Here's an interesting and probably unrelated tidbit: the temporary file name can get components from other files that you have open.
<JordanBrown>
I have "egg.scad" open in one tab, and in another new tab I get a backup file name of
<JordanBrown>
In fact, I get that even for files that *are* saved.
<JordanBrown>
And it's the same name for all of my tabs, which seems tacky.
<InPhase>
Hmm. I press space on a recently opened file, and it only creates an entry in backups IF Automatic Design and Preview is set. If it's disabled, it does NOT create a new file in backups at pressing space.
<InPhase>
So this does not trigger the preview, but is a consequence of it.
<JordanBrown>
yes
<JordanBrown>
yes
<JordanBrown>
I did say "probably unrelated".
<InPhase>
Well, that was my only operating theory. :)
<JordanBrown>
Well, OK, I have an interesting tidbit. It does *not* preview on keystrokes.
<JordanBrown>
(What?!? I hear you say.)
<JordanBrown>
It previews all the time, more or less continuously.
<JordanBrown>
Looks like it's probably every 200ms.
germ_ has quit [Ping timeout: 248 seconds]
peeps[zen] has joined #openscad
peepsalot has quit [Ping timeout: 276 seconds]
germ has joined #openscad
<JordanBrown>
OK, not quite. It *thinks* about previewing all the time. One might think that when it calls compile(), it would compile, but no, that only means that it's considering compiling.
<JordanBrown>
// If the file hasn't changed, we might still need to compile it
<JordanBrown>
// if we haven't yet compiled the current text.
<JordanBrown>
So it attempts to automatically preview on any change to the buffer.
<JordanBrown>
But it gets that wrong, because it checks by comparing the size of the text.
<JordanBrown>
If you make a change that doesn't change the size of the text - e.g. select one character, and then type another to replace it - then it does *not* preview.
<JordanBrown>
Preview on every internal-editor change seems like a bad idea.
<JordanBrown>
(Though I have to admit that it's kind of cool when I've got an animation running and it does that sort of automatically.)
LordOfBikes has quit [Read error: Connection reset by peer]
mmu_man has quit [Ping timeout: 272 seconds]
LordOfBikes has joined #openscad
J25k32 has quit [Quit: Client closed]
J25k32 has joined #openscad
kintel has joined #openscad
<kintel>
J25k32 Oh no. Don't tell me GitHub is making this a permanent change. That's actually one of the things which would make me switch
<kintel>
If that's the case, I'm sure competitors will be lining up with conversion offers :)
<J25k32>
kintel i don't have more info
<J25k32>
but microsoft is soooo bad
<kintel>
Perhaps it's a good thing - GitHub is becoming a bit too dominant
<J25k32>
if they need to do this because of scraper bots AI other will face similar issues
<J25k32>
but guess it is more like instagram so u can't really use it without login anymore
<kintel>
Following that logic, everything on the web will need logins to avoid bots scraping it to collect info : /
<J25k32>
yeah and having a login for a bot is also not that problem but sure can be tracked more easy
<kintel>
I guess we'll soon have to pay for the freedom to distribute out code freely
<J25k32>
and should be easy to just block if the same accesses 1000 repositories
<J25k32>
i think as each browser has a fingerprint there are for sure nicer solutions possible
<kintel>
yeah, I'd think rate limiting should only kick in if scraping a large amount of data, not for anything humans are able to do
<J25k32>
maybe having something like wikipedia for code
<InPhase>
kintel: It sounds like it only applies to "raw file" requests, and that they've already adjusted the rate limit upward to 20 in a few minutes.
<InPhase>
kintel: Some defined rate limit for unauthenticated downloads is fair and reasonable, as long as that rate limit is not impeding normal human access, and does not impede normal scripting workflows.
<kintel>
InPhase Yeah, let's see where it lands before jumping to conclusions :) I still see this as an early signal of how dependence on GitHub might be a risk
<InPhase>
kintel: Always has been. :)
<InPhase>
kintel: But at least they're listening and reacting.
<teepee>
ah, yes, I got that rate limit error yesterday too on iPad
<kintel>
I wonder if auto-mirroring is better these days. At some point I wanted to sync to GitLab, but mirroring was too messy to set up
<J25k32>
i mean you can't install windows without login anymore
<InPhase>
I try not to be first in line to defend anything that is technically owned by Microsoft. But it might just be a case here of a poorly chosen threshold.
<InPhase>
I think the secret to addressing bot scrapers is to set a larger threshold for a longer time period.
<teepee>
oh, we certainly can install our own git thingy, but I'd wait a bit for that :)
<J25k32>
kintel: file mirror should work but the workflow solutions are probably not 100% compatible
<kintel>
Could just be a classic "move fast and break things" engineering error. I'm not familiar with eng culture at GitHub to comment on that ;)
<teepee>
yep, for now I assume it's just some new knob they turned down too much
<kintel>
J25k32 Even file mirror wasn't doable back then, unless manually actuated
<InPhase>
Migrating the issues would be the challenge.
<teepee>
not really
<InPhase>
Especially without the matching accounts.
<teepee>
true, the users would be lost, but I do backup most stuff monthly
<J25k32>
kintel so you have always to push into multiple gits
<kintel>
Just mirroring repo, tags, branches, history and assets would be great. Mirroring workflow is hard.
<teepee>
the "migration" backup has everything including the issues
<teepee>
hmm, I need to check if the wiki is included, but that's a git repo anyway that can be cloned
<InPhase>
teepee: For all the repos under openscad?
<kintel>
Export is easy, but mirror on every action is hard - e.g. how to keep a GitLab version in sync with GitHub
<teepee>
not all, but maybe I should actually do all
<kintel>
I was initially looking into that to be able to e.g. run certain CI runners on GitLab