<samueldr>
what? firefox, while not in use, just now unloaded a bunch of extensions and the theme I chose, with a message (that now vanished) saying one or more extensions couldn't be verified
<samueldr>
if it lost all my settings for extensions, I'm really thinking going back to chromium, as much as I want to like firefox, it looks like it's trying hard for months to make me dislike it :/
<gchristensen>
maybe give it a second samueldr
<samueldr>
a second?
<gchristensen>
looks like it isn't just impacting you
<samueldr>
yeah
<samueldr>
if it really is outputting empty xpis, sure
<samueldr>
but that the sync system can just remove everything randomly, it's frankly spooky
<samueldr>
I'm not _yet_ throwing the towel, but it sure gives me thoughts :(
<gchristensen>
surely chrome's can too
drakonis has quit [Ping timeout: 276 seconds]
<samueldr>
chrome's behaviour in sync issues is generally to go the other way, if you bring up an older profile, it will re-sync extensions in an additive way in case of doubt, so extensions you removed might be re-added
<gchristensen>
this is not a preference error but a cryptographic validation error it seems?
drakonis has joined #nixos-chat
<samueldr>
?
<samueldr>
well, "not really"
<samueldr>
it errors since it's an empty file
<samueldr>
I don't see messages in mibbit :/
<gchristensen>
about that, it doesn't seem to be an empty file -- but a redirect
<samueldr>
oh
<samueldr>
right
<samueldr>
though, for me it seemed to have happend in a sync event
<samueldr>
and it seems to have really disabled extensions
<samueldr>
I hope that it only disabled, and not removed them outright
<gchristensen>
because I use containers for everything, and containers is a plugin
<samueldr>
oh, sorry
<samueldr>
at least I know it's now something I did :/
<samueldr>
also surreal how _some_ extensions are left working
<gchristensen>
I had 3, now 0
* samueldr
exports onetab data while it's still there
<gchristensen>
ok just got confirmation two peopleI'v pinged are trying to contact the right team
<samueldr>
(I'm on #firefox too)
<gchristensen>
cool
<gchristensen>
welp that channel rapidly went nowhere good
<samueldr>
oof, that brings the good side of people
<gchristensen>
haha yeah
aanderse has joined #nixos-chat
<aanderse>
so what do you use for bookmark+tab sync samueldr
<samueldr>
I don't
<aanderse>
oh the horror!
<samueldr>
I declared bookmarks bankruptcy
<samueldr>
it's where links go to die
<aanderse>
too many links at work to remember
<aanderse>
i couldn't possibly get by without some sort of bookmarking program
<aanderse>
ok but tab sync
<aanderse>
that not important?
<samueldr>
tabs are device-dependent
<aanderse>
oh boy not for me
<samueldr>
my personal computer doesn't care about my work computer tabs
<samueldr>
and I often cull them
<samueldr>
I hate having so many tabs open
<samueldr>
at the end it just becomes a graveyard like my bookmarks :)
<aanderse>
yeah i hate that too
<aanderse>
but
<ivan>
I keep my browser profile on an NFS mount and have the remote machine kill the other browser before starting
<samueldr>
I try to practice a kind of buddhism(?) by having zero tabs always :)
<aanderse>
i'll start looking at something on the commute to work
<aanderse>
then get to work
<aanderse>
and open the tab there
<ivan>
I hate the browser sync things for not syncing open tabs and cookies and visited links
drakonis1 has joined #nixos-chat
<samueldr>
that'd make it more useful to me
<samueldr>
simply tabs/bookmarks, not needed due to how I never end up really using them
<drakonis1>
finally, its the weekend
<drakonis1>
i'm so tired
<gchristensen>
sounds like a cert expired
<drakonis1>
are there any cool nix news?
<samueldr>
y'all, set xpinstall.signatures.required to false
<samueldr>
aanderse: gchristensen ^
<samueldr>
it just brought back the extensions directly
<samueldr>
literally instantly after toggling it
<samueldr>
so, good thing: extension data is still there
<aanderse>
samueldr++
<{^_^}>
samueldr's karma got increased to 85
<aanderse>
drakonis1: sounds like none today
<drakonis1>
oh well
<gchristensen>
nice
<drakonis1>
i'm preparing to bring up a nixos install soon.
<aanderse>
i would have liked to put some more time into cleaning up my nixops network today... but i had one of those days which just seemed to disappear
<aanderse>
spent 2-3 hours going through some horrible ORM db query which took ~10 seconds to execute :\
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 258 seconds]
Shados has joined #nixos-chat
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-chat
<lopsided98>
Firefox just disabled everybody's addons
<jasongrossman>
There was a discussion of the Firefox problem in this channel a few hours ago, including a solution.
<jasongrossman>
I guess I mean a workaround.
<lopsided98>
oh, I missed that
<jasongrossman>
Took me a minute to find it, although it turns out to be on this very page! (In my reader.)
<jasongrossman>
"<samueldr> y'all, set xpinstall.signatures.required to false [11:15]"
<samueldr>
infinisil: using firefox sync?
<gchristensen>
Toggling xpinstall.signatures.required seems to be a temporary workaround. (https://lobste.rs/s/u42xyr/firefox_disables_most_extensions#c_ljgdmh). This requires a build which is either not branded, or not considered “stable” Firefox ( nix-shell -p firefox-beta-bin ). They published a Discourse post as well for updates:
<clever>
pie_: i have seen some "closed source" minecraft mods, on github, and the gradle files are missing, so you cant build it :P
<pie_>
clever, we need a nix decompiler
<pie_>
but it decompiles projects into a build script
<pie_>
shhh its not supposed to make sense
<dtz>
doit do it
<dtz>
it doesn't need to make sense, it just needs to exist and work
<dtz>
;)
drakonis_ has quit [Ping timeout: 258 seconds]
drakonis_ has joined #nixos-chat
<jasongrossman>
pie_ Yeah - I feel like we should (really, not sarcastically) all be very excited about the potential of Nix as a cross-platform package manager. Just personally, I'm not, because I have no interest in proprietary OSes.
<jasongrossman>
But that's just me.
<pie_>
dtz, makng sense is a prerequisite of existence
<pie_>
theres things that dont make sense, and then theres things that dont make sense :P
<simpson>
jasongrossman: Nix already works on both Linux and Darwin, and there's many folks whispering about whether to try supporting more BSDs.
<jasongrossman>
simpson: Yes, that's very good IMO.
<dtz>
pie_: i want to live in that world. Happy to help build it! :)
<infinisil>
jasongrossman: It apparently also works pretty well on WSL
<infinisil>
But yeah, Nix as a cross-platform PM sounds great
<jasongrossman>
infinisil: Somehow I'd missed the news that it works on WSL. That is gerat news.
<jasongrossman>
infinisil++
<{^_^}>
infinisil's karma got increased to 95
<jasongrossman>
Sort of relatedly, the reason I'm not interested in guix-sd is that it won't ever include ZFS.
<jasongrossman>
(Actually, I'm just assuming that, given its philosophy. Maybe there's a workaround.)
<infinisil>
Oh I haven't thought of that, but that's a good point
<jasongrossman>
Yeah. Everything else about guix that I can see is either good or fixable, but not that.
<infinisil>
I guess it just won't be in the main repo, somebody else will have to maintain it
<jasongrossman>
For guix, right. For guixSD (which I think they've recently renamed to guix system or something - anyway, the equivalent of NixOS) I doubt it'll ever work well, even in a separate repo. I'd like to be wrong though.
<jasongrossman>
Oh I just said something dumb. ZFS is no use in guix if it's not in guixSD, is it?
<infinisil>
Well in Nix-land we have NUR <https://github.com/nix-community/NUR>, which is like a third-party package repo. They could gave something similar
<infinisil>
(well not only package repo, also can have overlays/nixos modules/any nix expression)
<jasongrossman>
Right, they could and should, but it's not going to work for OS-level things, is it? And it'll be the same with any other borderline-free things. I actually LIKE package managers that prefer free software, but the way guix is doing it makes it very hard for them to deal with edge cases.
<infinisil>
I don't know whether they have something equivalent to NixOS modules. Because NixOS modules allow you to do OS-level things (assuming the main repo is structured to allow for easy customization in that)
<infinisil>
You could easily add support for another file system in NixOS
<jasongrossman>
Even if they do, and even if someone makes one for guixSD, can you imagine switching to it if you have important files on a ZFS disk?!
<infinisil>
Yeah not sure
<jasongrossman>
The guix community is small enough as it is; a hypothetical ZFS-on-guix community is going to be tiny and therefore totally not reliable, no matter how great the individuals in it are.
<jasongrossman>
So you could lose access to all your files on ZFS. Eek.
<jasongrossman>
Whereas in NixOS it's part of the QA in each release.
<infinisil>
Eh if they just use stable versions and don't update it ever, it shouldn't lead to data loss
<infinisil>
Probably
<infinisil>
Ah yeah, testing and stuff..
<infinisil>
That's a good argument
<jasongrossman>
Not loss, right, just inability to carry on using the latest version of the OS. As often happens with MacOS and "third-party" software. Which is OK for minor things but not for a disk format, I should think.
<infinisil>
Well I guess you're not the target audience if you can't give up on non-free/libre software
<jasongrossman>
infinisil: That's what I mean, yes, BUT I'm pointing out that there are awkward borderline cases, so it's a problem even for people like me who only use free software but whose definitions don't quite exactly line up with the FSF's.
<jasongrossman>
And some of them are important cases.
<drakonis1>
one thing guix got going for them, is that they're cooperating with the software heritage project
<drakonis1>
it makes for really good advertisement
<jasongrossman>
drakonis1: Yes.
<drakonis1>
as for nix on bsd, i haven't heard anything regarding that in some time now.
<drakonis1>
i should ask the person who was trying to port it
<pie_>
i poke a guy sometimes but he hasnt answered yet
<drakonis1>
the other thing guix got going is that they have an installer now
<drakonis1>
so its currently far more approachable than nix
<drakonis1>
nixos and guix system rather.
<pie_>
drakonis1, hm
<drakonis1>
their 1.0 blog is fairly good at selling guix
<drakonis1>
there's also a increasing amount of packages
<jasongrossman>
An installer is good, yes.
<jasongrossman>
There's been a lot of "someone should do that" here. But the someones are not only busy but also doing super productive things for Nix. :-)
<pie_>
i know two guys that were considering doing it and then im not sure what happened :D
<pie_>
whoa
<pie_>
main chan has 900 users in it
<pie_>
thats a big irc channel
<drakonis1>
lots of idlers though
<drakonis1>
there's us the regulars and there's the idlers
<drakonis1>
maybe something nix could enjoy is achiving lisp levels of hackability
<drakonis1>
but then that would just make it lisp wouldnt it
<infinisil>
jasongrossman: I think aszlig's nixpart goes a bit in that direction, regarding the partitioning at least
<infinisil>
Well yeah, but nobody wants to write that lol
<infinisil>
For an installer usecase i mean
<pie_>
i figued but still
<drakonis1>
why don't we sign up for gsoc?
<dtz>
b/c we aren't organized enough :P
<dtz>
err, sorry. I don't know! I think a lot of us would love to help witht hat.
<drakonis1>
gsod came up not a week ago
<drakonis1>
could've signed up for that
<samueldr>
people are busy, those things take time from more than one person usually, to organize
<samueldr>
99% sure the team being mostly stretched time-wise is the main reason
<drakonis1>
fair enough
<samueldr>
for GSOD, it requires two organization admins, plus the mentors, to commit time
drakonis1 has quit [Quit: WeeChat 2.4]
jtojnar has quit [Remote host closed the connection]
joepie91 has joined #nixos-chat
endformationage has quit [Ping timeout: 250 seconds]
hedning_ has quit [Remote host closed the connection]
__monty__ has joined #nixos-chat
<gchristensen>
seems a fix is out for the browsner thing
<gchristensen>
there was an effort between Domen + Hercules, and gilligan + Tweag to get GSOD going, but something happened and we ran out of time
colemickens_ has quit [Quit: Connection closed for inactivity]
jasongrossman has quit [Read error: Connection reset by peer]
<infinisil>
Trying to get some sleep, have the window open, hear how it rains softly
<infinisil>
And now it just turned into a hail-storm!
<infinisil>
Damnit
Peetz0r has quit [Ping timeout: 264 seconds]
<__monty__>
Don't sleep so late, infinisil : p
<infinisil>
The night is my most productive time though!
Peetz0r has joined #nixos-chat
<eyJhb>
Update on my Macine Learning which turned out to be a RNG, it actually works now (as intended in recognizing road signs)!
jasongrossman has joined #nixos-chat
<MichaelRaskin>
All ML is RNG, the question is whether you cen ensure enough bias in the desired direction…
<joepie91>
rejoice! I have solved a parallelism issue in my JS streams implementation, and accidentally solved another design issue that has been bothering me for like 2 weeks now
<joepie91>
and now I can supported pull-based streams with unlimited concurrency :D
<joepie91>
support*
<MichaelRaskin>
Congratulations
<joepie91>
so the original issue was that it was not possible to efficiently represent something like "a stream of TCP clients" in the model, because you could only parallelize the upstream part of the pipeline with a fixed number of slots, because (by design) a stream cannot know what the upstream streams are doing or how many items there are left to process
<joepie91>
but of course you often want to just handle incoming TCP clients as they come in, not "8 at a time"
<joepie91>
the issue from 2 weeks ago was about 'looped pipelines' and parallelism; ie. a pipeline where the source stream can be 'refilled' by a downstream stream, eg. in the case of a task queue represented as a stream
<joepie91>
in which case, if you parallelize reads, and thus you read from the queue while other reads are still in-flight, you can't reliably determine whether the queue is temporarily exhausted (and will eventually be refilled by an in-flight read further downstream), or permanently exhausted (and therefore reached end-of-stream)
<joepie91>
and since streams don't know about downstream *either* (again, by design), it can't know when all reads have finished processing fully
<joepie91>
turns out that both of these can be reasonably efficiently expressed through a 'peek' API that simply returns "true" if any upstream stream has values in its buffer
<joepie91>
and some logic in parallelization streams that force it into sequential mode once it gets a "no more buffered values available" signal, and that goes back to parallel mode once it gets a successful value again
<joepie91>
</ramble>
<joepie91>
anyway, I'll probably write a blogpost with the details and examples etc. at some point, because most of the above is probably incomprehensible unless you've worked with pull-based stream abstractions :P
<joepie91>
I'm just Really Excited to have finally solved these blockers
<manveru>
first time i hear of js streams...
<manveru>
guess i haven't done enough node to stumble over that term
<joepie91>
manveru: there are various stream abstractions; Node streams are the most common one though they're junk, pull-streams also exist but haven't really kept up with recent changes to JS and are not very well-specified
<joepie91>
it's not JS-specific though, futures.rs (in Rust) has streams too, same concept
<joepie91>
anyway, I'm working on a derivative of pull-streams, that actually works with things like promises and that has well-specified behaviour
<manveru>
so it's like eventmachine?
<joepie91>
eventmachine is just an event loop, isn't it?
<manveru>
mostly
__monty__ has quit [Quit: leaving]
<manveru>
but also provides non-blocking i/o
<joepie91>
then eventmachine is more equivalent to JS/Node
<joepie91>
this streams thing is built on top of that
<joepie91>
manveru: anyway, like an array of values is basically values distributed over space, a *stream* of values is values distributed over time
<joepie91>
ie. it's an abstract source of values over time
<joepie91>
and with a good streams/pipeline abstraction, you can fairly easily specify a series of operations to apply to a stream-of-values-over-time
<joepie91>
which is useful for a number of things; processing large files without loading it all into memory at once; as an event-handling abstraction (eg. handling a stream of TCP clients), and so on
<joepie91>
anything where you repeatedly apply operations of some sort over same-shaped values that come in over time, really
<joepie91>
anyway! the core streams implementation that ships with Node is... not very well-designed, and really awkward to work with
<joepie91>
hence the need for a better design
waleee has joined #nixos-chat
waleee has quit [Ping timeout: 252 seconds]
waleee has joined #nixos-chat
jonge has quit [Quit: quit]
endformationage has joined #nixos-chat
jtojnar has joined #nixos-chat
waleee has quit [Ping timeout: 246 seconds]
<gchristensen>
since I lost all my FF containers, I'm considering containerizing via many profiles instead.... MichaelRaskin you do this yeah?
<tilpner>
Huh, it forgot which containers a tab belonged to?
<tilpner>
My addons seem to have kept their state after enabling them again
<gchristensen>
all my addons did, except Multiple Containers
<gchristensen>
that one reverted to the 4 default containers it ships with
<MichaelRaskin>
Well, I use transient profiles
<tilpner>
:/
<gchristensen>
(I had 50-75 containers setup)
<MichaelRaskin>
Prebuilt single-start profile that I clone and run Firefox in nsjail
<tilpner>
zfs rollback to when it had the containers associated, then disable signature verification without starting FF?
<gchristensen>
interesting. how do you handle logins, MichaelRaskin?
<MichaelRaskin>
Password manager?
<gchristensen>
you don't stay persistently logged in?
<MichaelRaskin>
Well, I have some Firefox instances that run for days
<MichaelRaskin>
But if I reboot, I relogin
pie_ has left #nixos-chat ["Leaving"]
pie_ has joined #nixos-chat
<pie_>
gchristensen, ouch
<gchristensen>
interesting
<MichaelRaskin>
It wouldn't be hard to have persistent profiles, actually
<gchristensen>
how did you construct the initial profile?
<MichaelRaskin>
The same jailing setup is used for desktop Signal; there I do have persistent profile (but it is not a generic browser)
<MichaelRaskin>
Well, it is a moderately crazy expression
<MichaelRaskin>
It doesn't use much of my system setup (it's a normal Nix expression)
<MichaelRaskin>
(but this assumes some nsjail wrapping)
<gchristensen>
this is amazing!
<MichaelRaskin>
This doesn't include extension installation; maybe I will add it once I understand what Nix repo tracks the upstream XPIs reasonably well
<gchristensen>
that does feel like an important part
<MichaelRaskin>
Well, you might notice it includes Marionette…
<pie_>
i was having some trouble figuring this stuff out at some point :p
<MichaelRaskin>
Some features normally controlled via extensions are controlled via Marionette
<gchristensen>
like which?
<MichaelRaskin>
JS on/off toggling in runtime (new instances are JS-off by default, it is easy to launch a JS-on instance, and it is easy to control JS-on/off externally)
<MichaelRaskin>
pie_: my approach still doesn't solve the damned locked preference entries
<pie_>
MichaelRaskin, what do you mean
<MichaelRaskin>
There are entries one cannot change via prefs.js or about:config
<pie_>
oh
<pie_>
huh
<pie_>
why not? do you have an example
<MichaelRaskin>
And if profiles are disposable, I don't care about updating profiles
<pie_>
" This controls a diminishing number of things and is intended to be deprecated, and then removed." though
waleee has joined #nixos-chat
<gchristensen>
ehh this is not what I want to be doing :P
<MichaelRaskin>
The real answer is of course that I generally decrease the amount of stuff I do via GUI browser at all
<MichaelRaskin>
(so extensions are less useful for me)
<gchristensen>
yeah, right. my extensions are: multiple containers, ublock origin, and passff
<pie_>
yeah now its just electron everywhere :p
<MichaelRaskin>
Right, containers are completely irrelevant for my setup, I block some completely outrageous domain via Squid ACL (and lack of scripts also decreases the need), and I definitely use external control for password management
<gchristensen>
external control, ie: your browser can't automatically pull in passwords?
<pie_>
anyway you guys sound like you have some cool setups
<MichaelRaskin>
The only Electron garbage pile I use is Signal
<gchristensen>
this is all MichaelRaskin's cool setup :P my setup is boring
<gchristensen>
every person and company is doomed to let a cert expire from time to time
<gchristensen>
it is the law
<pie_>
im just feeling negative sentiment in general
drakonis_ has joined #nixos-chat
<pie_>
i can completely understand mistakes, things happen, things get forgotten
<pie_>
has mozilla made any *good* moves recently?
drakonis has quit [Ping timeout: 264 seconds]
<gchristensen>
I mean
<samueldr>
good moves are not good for the news cycle ad prints
<gchristensen>
yeah all their Rust work is paying off big time
<samueldr>
otherwise you'd hear more good moves from google and apple and microsoft and mozilla and and and
drakonis has joined #nixos-chat
<pie_>
samueldr, yeah :/
<samueldr>
the way "communication" (for lack of a better term) is done these days and age, mostly things that are outrageous gets shared around
<samueldr>
that's not specific to this situation
<samueldr>
when was the last time you remember about a terrifically good thing being shared all around, instead of being a targetted focus?
<samueldr>
like sure, I'll know about $good from $x since I follow $x news
<pie_>
yeah
<samueldr>
but if $y does anything bad, everyone, their cat and their fleas will share it
<samueldr>
and I hate that
<pie_>
MichaelRaskin, sidenote, why *doesnt* firefox just have a "disable all telemetry and similar" button - its supposed to be The Good Browser
drakonis_ has quit [Ping timeout: 248 seconds]
<MichaelRaskin>
Well, I would say what qualifies Mozilla story for controlled flight into ground is that they were _told_ to allow user control over extension signing…
<pie_>
i thikn i read somewhere that debian actually implemented something for that
<pie_>
but im lost between all my links right now
<samueldr>
I have this unpopular opinion, be ready to downvote me, but while Mozilla is possibly less bad, it's a posturing thing and actions aren't really there to back it imho :/
jtojnar has quit [Quit: jtojnar]
* gchristensen
downvotes
<MichaelRaskin>
I would say that _Google's_ actions are there to make sure that Mozilla is way less bad
<samueldr>
MichaelRaskin: an appropriate rehash of my opinion
<MichaelRaskin>
Mozilla is an interaction of UI designers and marketers with people who actually understand technical decisions and verifiability of promises
<MichaelRaskin>
They have some complicated decision making procedure that is quite unlikely to le through Google-scale evil moves, but can do a lot of cluelessness
<pie_>
samueldr, i dont have anything to back it up but my general sentiment is strong agreement
<MichaelRaskin>
There are many people who are clueless enough not to understand that their actions do not fit the proclaimed goals
Myhlamaeus has quit [Ping timeout: 250 seconds]
* pie_
mumbles to self, think about rust think about rust
<MichaelRaskin>
Why Rust is good: because there the marketers just don't have any strong opinion
<pie_>
MichaelRaskin, or it hasnt been popular long enough :p
<MichaelRaskin>
Remember forever the words of Linux Torvalds in the kernel's coding style guide: «not evil, just severely misguided»
<samueldr>
part of what coloured my opinions vis-à-vis mozilla: the whole web extensions removing what makes µblock work right... no mozilla developers openly critiqued or shot down the idea. Mozilla (AFAIUI) _are_ in the decision making pipeline, and have already said that they will implement the new web extensions spec update
<samueldr>
speaking of, no news about that situation, so I guess either nothing new happened yet, or a positive resolution happend (thus no news cycle :/)
<MichaelRaskin>
I think 66.0.4 is not yet released
<samueldr>
obviously, yersterday's issue now peppers search results and makes it impossible to fine what I had in mind :/
<MichaelRaskin>
Well, we know that Google deliberately break the parts of WHATWG HTML5 LS that they previously had working
<MichaelRaskin>
Therefore Chromium will not have working blockers, whatever happens with the spec
<MichaelRaskin>
Mozilla will implement whatever «streamlined» version Google implements, because portability (and reduced version is still sometimes useful)
<samueldr>
oh, finally, in firefox 67 mozilla will disallow extensions to run in incognito, yay
<MichaelRaskin>
Ouch
<MichaelRaskin>
Of course, I personally don't really care…
<samueldr>
allow extensions to be disallowed*
<samueldr>
or the other way around, you can enable extensions in incognito extension-by-extension
<MichaelRaskin>
Ah, wait, that's a positive
<samueldr>
yes
<samueldr>
something chrome had since they had extensions
<samueldr>
and a net minus in comparison
<flokli>
samueldr: keeping ublock inside icognito mode would still be nice :-)
<samueldr>
flokli: then allow it :)
<flokli>
I guess I misunderstood. So it's still possible? good :-)
<samueldr>
in fact, I find this useful _because_ I only enable µblock, and nothing else, so no complex µmatrix blocking, easier to deal with a one-time reading of a blog that lights up µmatrix like a christmas tree
<MichaelRaskin>
(re: APIs for uBlock — Mozilla has a ton of extra WebExtrnsion-like APIs anyway, so they will probably keep the working version, too)
<samueldr>
yes, the update coming in 67, if I read it right, seems like a net positive
<MichaelRaskin>
I sincerely believe that most of the web is strictly better without JS _and_ CSS
<samueldr>
my opinion is that there are two "webs" that are in conflict in their base needs
<samueldr>
(1) content (2) apps
<MichaelRaskin>
Yeah, that's true.
<samueldr>
and (2) probably shouldn't have been built off of (1)
<MichaelRaskin>
On the other hand, «one-off that lights up uMatrix» is usually the type (1)
<samueldr>
enabled by techs from (2) :)
<MichaelRaskin>
Doesn't matter, block everything including inline styles
<samueldr>
you're the 0.1% for which I still develop fore :D
<MichaelRaskin>
What is fore?
<samueldr>
typo'd for
<samueldr>
when I do web, the styles are applied to a tree that makes sense without styles
<samueldr>
and (when not constrained by demands) the js parts are enhancements and shouldn't leave a site non-working without it working
<samueldr>
you know, basic web development skills :/
<samueldr>
(sorry, almost getting to the "get off my lawn" part of the rant about web)
<MichaelRaskin>
Is GeckoView available on desktop, by the way?
<samueldr>
I was wondering the same earlier today
<MichaelRaskin>
Apparently no
<pie_>
samueldr, ++
<pie_>
though i never tries disabling css
<pie_>
tried
<pie_>
so should i start running my firefox through a vm with everything going through a proxy and noscript-like whitelisting
<pie_>
but on the proxy :p
<samueldr>
many app-like websites will pepper the DOM with things semantically unfit to live in the DOM, and contents website now often follow app-like ways to build a website, so they're even worse most of the time
<MichaelRaskin>
I am more annoyef that Gitlab doesn't allow _viewing_ without Js
<samueldr>
(js is still active, this combination sometimes breaks more assumptions)
<pie_>
MichaelRaskin, oh god thats the most hell annoying thing ever
<samueldr>
gitlab's UX is extremely bad, with too much lazy-loaded bits, which will often load way too late and make the page jump around
* samueldr
grumps
<joepie91>
samueldr: nono, you're getting it all wrong, it has excellent time-to-paint!
<joepie91>
:^)
<samueldr>
about:blank has as much useful information and better time to paint :D
<andi->
turning my computer off is also a very smooth black screen experience :D
<samueldr>
time-to-not-paint == 0
<MichaelRaskin>
Conspirologic theory of Mozilla story: someone wanted to remind technically competent people just _how_ horrible the web is without adblocker
<pie_>
heh
<pie_>
thats a pretty positive outlook for a conspiracy though
<joepie91>
one has got to be my kind of crazy to implement something like that
<infinisil>
Definitely
<gchristensen>
reimplementing ZFS in any fashion seems a unique brand of fascinating
<joepie91>
hehe
<MichaelRaskin>
Hm, will I succeed in testing nix-quote-urls without rebooting…
<pie_>
MichaelRaskin, vm? :P
<MichaelRaskin>
That's not the problem
<MichaelRaskin>
Well, I can just grow the logical volume… I guess
<MichaelRaskin>
But maybe it won't take _all_ the remaining /nix/
<pie_>
oh lol rebuilding the worl?
<MichaelRaskin>
(first attempt failed on running out of space, GC didn't free much, as last-booted and the current are a mass-rebuild away from each other — but it did free 30GB, should be enough, right?)
<MichaelRaskin>
I am using the default.nix which is pinned to yet another Nixpkgs revision, apparently a mass rebuild away from _both_ my current and my booted…
<infinisil>
Testing it on nixpkgs with `nix-quote-urls **/*.nix` finishes in 27 seconds for me :)
<MichaelRaskin>
I should write nix-fmt in Common Lisp, compile it with ECL and say to everyone that this is less closure overhead than Boost anyway
<pie_>
MichaelRaskin, :D
<MichaelRaskin>
Does Mozilla fund optimisation work for GHC? To make Rust compile times look better in comparison
<MichaelRaskin>
Apparently, enough people did to make Razer reconsider whether it was an April 1 joke or just a pre-announcement
<pie_>
maybe someone at razer just really wanted a toaster that didnt suck
<pie_>
not sure id want my toaster to look like an old xbox
<MichaelRaskin>
infinisil: come on, in Common Lisp building a PEG parser, re-parsing PEG definition, parsing a PEG for BNF, parsing BNF for TPTP, converting it to PEG, and compiling this _still_ takes less than just compiling hnix!
<joepie91>
not with the lever design they have now
<pie_>
ok fine
<pie_>
:P
<joepie91>
though in fairness, props to Razer for actually going through with this lol
<joepie91>
that has *got* to be an internal joke that eventually got promoted to "eh, well, fuck it, why not"
<pie_>
joepie91, yeah somehow i had the nagging idea that household appliances these days totally suck and quality isnt actually improving, which is super ironic but
<pie_>
aaah now im angry at society again
<joepie91>
lol
<joepie91>
there's definitely improvements
<joepie91>
just not exclusively...
<pie_>
joepie91, this toaster is the best thing since sliced bread
<infinisil>
MichaelRaskin: Hehe, quite possibly
<MichaelRaskin>
Wait what. It builds hnix _again_
<joepie91>
pie_: ehhehehehehe
<infinisil>
MichaelRaskin: Ah yeah I guess I could've turned that off
<infinisil>
MichaelRaskin: It does profiling builds as well by default
<infinisil>
But damn, you have a slow machine
<MichaelRaskin>
It's a laptop
<MichaelRaskin>
Ah right, I didn't enable using more than one core
<MichaelRaskin>
I guess if we want this in ofBorg, filtering to run only on the changed files will be necessary
<pie_>
joepie91, ok that toaster is actually amaying
<pie_>
using the heat expansion to lower the bread
<pie_>
that means your bread wont be stuck in case of power loss or something
<gchristensen>
wait, the toaster isn't a weird reference to their computers overheating?
<MichaelRaskin>
It probably was three years ago
<infinisil>
MichaelRaskin: Yeah, I guess for ofborg I should add a mode that only checks for unquoted urls and errors if there are
<gchristensen>
yes please
<gchristensen>
if you could emit a JSON blob of exactly where (file, position) I can add in-line suggestions
<MichaelRaskin>
I can write a shell script that does that, just on the files different from the specified branch
<infinisil>
Ah yeah that would be nice
<infinisil>
Not today though
<MichaelRaskin>
The situation with my documentation PR is annoying. I didn't notice that the package-specific notes include some _user_ notes (as opposed to developer notes), so I added a small chapter for user notes. And I wrote it in markdown. Now I need to decide if I want to insist on splitting the package-specific notes, and I need to convert some of the notes between markdown and docbook…
<gchristensen>
yay docbook
<MichaelRaskin>
I am more likely to convert in the opposite direction as I want to keep the user notes higher in the manual…
<MichaelRaskin>
And DocBook is a complicated schema and we don't have good tools for troubleshooting faults in structuring DocBook
<pie_>
someone had some WYSIWYM editor
<samueldr>
if you're speaking about me, no, initial thoughts and research about ways to make one :(
<pie_>
oh. i could have sworn there were screenshots, maybe that was just a general xml editor thing