mcdanlj

mcdanlj, 1 day ago

@bricks Have you tried chucking short pieces of filament in a Dremel or other rotary tool and friction welding over the seam?

I haven't done that for seam hiding but have successfully used it for repair, and it might be easier to control than a pen.

mcdanlj, 4 days ago

@danderson I just learned how to fix that with a workaround, and what they are trying to do to fix it long-term.

https://social.makerforums.info/@mcdanlj/112372504552067471

mcdanlj, 4 days ago

@adamw Thank you! I'm grateful for the immutable/atomic work, and I've been through some sanding off rough edges before... It's worth a few of those for the benefits. ☺

mcdanlj, 4 days ago

@notting @adamw I understand why flatpak means "start your config over from scratch or move files around" but it's still a pain. I'd really like it if flatpak had a declarative way to offer to auto-migrate existing configuration. That might make me take it more seriously.

mcdanlj, 4 days ago

@garrett @notting @adamw Right, I can figure this out on a case by case basis. But it's not something that's reasonable to ask of all users generally. If part of the flatpak spec was where the configuration lives outside flatpak, then flatpak itself could offer to move (or copy) configuration when installing a flatpak for which existing "legacy" configuration existed.

It's somewhat analogous from an end user experience point of view to how browsers offer to import your data from other browsers. But done instead across locations for the same app, and configured declaratively so that it's implemented once in flatpak itself. The app itself literally can't do that because the whole point of the restriction is the additional sandboxing, as I understand it...

mcdanlj, 3 days ago

@siosm @adamw @garrett @notting my expectation wouldn't be moving just one profile. It would be the entire configuration. That by it declaring a set of one or more more (source, destination) pairs of paths, when installing any flatpak, you would be given the option of moving the config, and possibly on uninstall the option of moving the config back. This would be for any package that declared these pairs, and not specific to Firefox.

mcdanlj, 2 days ago

@siosm @adamw @garrett @notting I would expect something like that:

• it would by default be user-controlled and prompted
• if the target did not exist, the default would be to copy from the source to the target
• if the target already existed and was newer than the source, the default choice would be not to copy the files, and a choice might not be presented
• if the target already existed and was older than the source, the default could be to copy over it, with explicit permission from the user

This would reasonably handle both a migration from native to flatpak and simultaneous usage, and not require users to search for information on configuration files to have a seamless move.

mcdanlj, 2 days ago

@siosm @swick @adamw @garrett @notting It's definitely not only Firefox. I first ran into this with FreeCAD, as it happened. But it could easily have been anything. Almost everything stores configuration. Why would this be specific to Firefox?

mcdanlj, 2 days ago

@siosm @swick @adamw @garrett @notting To avoid mistakes on my part, where precisely on GitHub?

mcdanlj, 2 days ago

@siosm @swick @adamw @garrett @notting

it's currently installed in the base image in Fedora Atomic Desktops

That completely ignores everyone who is bringing along a home directory and is following widespread advice to change from native to flatpak.

mcdanlj, 2 days ago

@swick @siosm @adamw @garrett @notting I would suggest this only when a flatpak app is installed or uninstalled, thus the "source" and "target" language I used. (You might reasonably also suggest a separate capability for users to choose to initiate this at any time if they so desire, but that's not part of what I have been suggesting.)

If the config doesn't make sense to copy for some particular app, then it would be a bug to declare such an equivalence using this suggested feature. But this is not a normal case.

It is not complexly specified, so complexity would be an unnecessary aspect of implementation.

Calling this "little gain" and crudely dismissing it is discouragingly dismissive, and devalues the work that users have done in developing their preferred configuration. ☹️

If it were declared, then at the very least, users could get information in a consistent way about how to preserve their configuration, and not have to hope that they found the right stack overflow answer when they do a web search. The question of whether to automate copying could reasonably be a separate conversation from making a consistent record of equivalent paths for configuration. But the possibility of such a feature could add rationale for implementing the record in the first place.

mcdanlj, 1 day ago

@swick @siosm @adamw @garrett @notting Let's make perfect the enemy of better shall we? /s

FreeCAD has plenty of paths in configuration. Some of them might not work and still need fixing. It would still be better than starting over entirely from scratch, speaking as a user.

Please feel free to describe specific examples where you feel that this would be damaging so we can have a conversation about specifics rather than sweeping generalities.

mcdanlj, 1 day ago

@siosm @swick @adamw @garrett @notting I'm not ignoring this, by the way — my idea is to use this thread of conversation to think through and do a better job of writing up a more coherent feature suggestion.

mcdanlj, 4 days ago

@danderson I'm (happily) running Silverblue on work and home laptops, and I've certainly used tools starting with audit2allow to build a custom policy. What can't you find?

mcdanlj, 7 days ago

@FreeCAD I tried multiple times to submit an entry and couldn't tell whether it "took."

I don't know how to check whether an entry was submitted.

So I'm kind of stuck here.

mcdanlj, 7 days ago

@M4x @FreeCAD The fact that I never saw that doesn't mean the server didn't register it...

mcdanlj, 6 days ago

@FreeCAD I tried again, still with Firefox, and it still hung and gave up.

I finally gave up and used Chrome, and it worked.

If I have lots of votes in the system, it's not my fault.

Seems sad to me that the FreeCAD logo vote site doesn't work with Firefox, at least for me. 😢

mcdanlj, 10 days ago

@ldodds I'm using https://elk.zone as a PWA as my client on android, and it supports both multiple accounts and long-press to copy links.

As of last I checked, it doesn't work quite right as a PWA using mobile firefox, so it's basically my only use of Chrome on the phone... It has been quite a while, so maybe I should try again as a firefox PWA.

mcdanlj, 11 days ago

After being inflicted with this dubious feature for months, I've wondered whether Product Management in Google's Android group has a problem with child labor laws... 🙄

mcdanlj, 7 days ago

@pro @OpenSCAD Saw your post https://zaitcev.livejournal.com/266803.html and got back here — do you know that FreeCAD can render OpenSCAD and exports STEP? It uses OpenSCAD to more or less compile to CSG and then renders CSG.

Not suggesting that you shouldn't add STEP to OpenSCAD! ☺

mcdanlj, 6 days ago

@pro @OpenSCAD Your livejournal is in my RSS feed. ☺

I feel moderately fluent in FreeCAD by now, and what you describe seems very simple and quick to me; would be a bit faster for me in FreeCAD than in OpenSCAD. But not trying to talk you into a switch, and direct STEP output from OpenSCAD seems like an awesome addition.

Extrude a bar, pocket a V-notch, multi-transform to make a linear copy and then mirror it. (There are other ways depending on how it's intended to be arranged, I imagined it centered.) Took less time to model than to write the this description of how I did it... ☺

mcdanlj, 5 days ago

@pro @OpenSCAD Ah, that would have been even simpler in the design tree, just putting the three notches into a single sketch to pocket, and dropping the multi-transform, which was there just as an example for modeling with less drawing in a sketch.

I'm curious about a few things...

Are you assuming mm as the base unit when writing a STEP (current typical practice in OpenSCAD AFAIK) or are you including some way of explicitly defining the base unit?

Are you preserving color and representing it in the STEP?

When (on livejournal) you say "possibly square or round" do you mean you are representing a rectangular face as a single square instead of as two triangles as it would be in an STL, and stitching curves back together at some epsilon (like the "arc welder" plugin for 3d printing slicers)?

mcdanlj, 5 days ago

@pro I definitely have run into plenty of models that have to be scaled by 24.5x to make sense, and a few 10x, but fortunately, mm seems to have mostly won by now.

I've definitely worked successfully on STLs in FreeCAD, and I'm happy to share more, but not trying to talk you into it. ☺ So let me know if you want me to share more. Otherwise I'll just celebrate the idea of STEP export from OpenSCAD. 😀