|
Cocoa Crispies posted:the former yeah I guess that's true. biometric auth is dumb as hell
|
#
?
Dec 3, 2013 02:59
|
|
|
|
| # ? Jun 1, 2024 07:58 |
|
|
Suspicious Dish posted:we agree on this. of course defaults should be good. unfortunately the terminal is horribly bad as a ui. i think we want to make it accessible, but not confuse people who launch it by accident. every default keybinding we've tried to assign to it has pissed some people off. it used to be ctrl+shift+t, which is "undo close tab" in firefox, which confuses a bunch of people who come from windows. it then switched to ctrl+alt+shift+t, but that caused some compiz people to get mad, so then we tried super+t/win+t but people complained that it wasn't ergonomic, and that we should change it to something else. defaulting to "no keybinding" isnt the best solution, but its the best we've found. sounds like you've got terminal indecision (heh) is your goal to make an UI that will never confuse anyone in any situation? have you put numbers to "how many users do we have that will be confused by accidentally opening a terminal" versus "how many users do we have that use terminals regularly and would like any shortcut at all so they can sit down at a Redhat machine and be immediately productive"? if you want to make a business OS for people who work on computers, not with computers you shouldn't be loving around with making it work out-of-the-box for techies
|
|
#
?
Dec 3, 2013 03:52
|
|
|
suffix posted:if you want to make a business OS for people who work on computers, not with computers you shouldn't be loving around with making it work out-of-the-box for techies this is a core problem with gnome 3: who the gently caress is their audience? every unix user i have ever known is one of three things:
which one of these core user bases is demanding a desktop environment optimized for 1280x720 laptops with powerful 3d acceleration and such terrible screen area that there's no room for a window list or conventional start menu? it seems ilke the "aunt tillie" / "silent majority" poo poo, where they design for some imagined "casual" user instead of the real-life sperglords who need their apps to actually loving work
|
|
#
?
Dec 3, 2013 03:56
|
|
|
you know what really helps out with systems administration and unix software development? a file browser optimized for people who don't know what "files" are
|
|
#
?
Dec 3, 2013 03:59
|
|
|
 i'm the inconsistent font and style on different ui elements
|
|
#
?
Dec 3, 2013 04:05
|
|
|
i don't use gnome 3 and i'm not qualified to work on ui so don't take this personally but c'mon
|
|
#
?
Dec 3, 2013 04:06
|
|
|
serewit posted:i don't use gnome 3 and i'm not qualified to work on ui so don't take this personally but c'mon no one should take it personally suspicious dish is an ok poster and probably a good person who just happens to have a job working on a lovely product directed by crazy people. we've all worked on lovely products before
|
|
#
?
Dec 3, 2013 04:07
|
|
|
not sure where you got that image, because everything is configured by default to have consistent fonts. somebody had to mess around to change the titlebar font.
|
|
#
?
Dec 3, 2013 04:08
|
|
|
Suspicious Dish posted:not sure where you got that image, because everything is configured by default to have consistent fonts. somebody had to mess around to change the titlebar font. http://www.gnome.org/wp-content/uploads/2011/04/eog-screenshot.png
|
|
#
?
Dec 3, 2013 04:13
|
|
|
Notorious b.s.d. posted:suspicious dish is an ok poster and probably a good person who just happens to have a job working on a lovely product directed by crazy people. we've all worked on lovely products before so, gonna be absolutely honest here. i love my job. i work on kick-rear end open source projects with cool tech and amazing people and i don't have much overhead. there's still internal politics (which is absolutely boring to everybody else), but less than at other companies i've been told, and the management is taking huge steps towards fixing this. i have my own series of gripes with gnome3, because so does everybody. i do think it's going good places, but obviously as you've realized it's not ready yet. i think that mistakes were made when releasing gnome3, and we should have continued gnome2 development for a while before releasing gnome3. of course gnome has a giant problem with 'what is our audience' -- part of this problem is that gnome isn't controlled by any one company, and so it has no formal leadership. people use gnome tech everywhere, and we sort of have to listen to all the people who use it as a cheap windows replacement for govt/schools, the people that install it in the back of an airplane seat to watch movies on, and on those crazy linux people that install it on their mom's computer. i ask "how can we make gnome3 better for you" not because i don't know all of its shortcomings, but because i genuinely do want to improve the product and make sure people's feedback is heard. if anything, hopefully my posting in here has made you think about the problem space before and that some of these problems are more difficult than they first appear. i think the gnome team also has a giant fault in that it tends to be a bit too visionary -- we have lots of far-future plans where everything is magical and it pulls from google automatically, hooks up to steam/netflix, etc., and we design our features with this in mind, rather than the day-to-day life of people today. this is one reason why the background panel sucks: it's designed for this future where people chose their background from flickr/google drive instead of their filesystem, because that's often where photos like this are hosted. also, yes, i talked to my colleagues about all the issues people brought up, and of the ones we think are problems, we're working on fixing them.
|
|
#
?
Dec 3, 2013 04:18
|
|
|
funroll loops posted:hats are for unit vectors and estimates what's so hard about xbias * yregularization weight
|
|
#
?
Dec 3, 2013 04:42
|
|
|
Suspicious Dish posted:i think the gnome team also has a giant fault in that it tends to be a bit too visionary -- we have lots of far-future plans where everything is magical and it pulls from google automatically, hooks up to steam/netflix, etc., and we design our features with this in mind, rather than the day-to-day life of people today. this is one reason why the background panel sucks: it's designed for this future where people chose their background from flickr/google drive instead of their filesystem, because that's often where photos like this are hosted. look mang, i liked your earnest posting enough to try your stuff, but you know in your heart that this is where the crazy part applies. please make a ticket labeled "YAGNI" and close all "visionary" feature requests as dupes.
|
|
#
?
Dec 3, 2013 05:45
|
|
|
Share Bear posted:what's so hard about it's not hard and a lot of times you will see roman (i.e., non-italic) script in subscripts, though i'd probably do something like yweight or yreg depending on context also suspicious dish i see you a lot in grey forums and i think you're a good poster 
|
|
#
?
Dec 3, 2013 05:53
|
|
|
Brain Candy posted:look mang, i liked your earnest posting enough to try your stuff, but you know in your heart that this is where the crazy part applies. of course i know. this is a big flaw we have, and it's certainly one we're having trouble with. i bring it up internally all the time. at the same time, visionary stuff like this is what got the desktop where it is. we had to deal with lots of low-level plumbing before so that steam could run itself on a solid platform without crashing: networkmanager, pulseaudio, systemd, bluez, upower, accountsservice, and dri/aiglx were all invented inside the gnome community, and they're the things that allow valve to consider building steamos. you've probably had issues with all of these if you've ever dealt with them, and we're sorry, but things like this are complex and it took years for us to iron out the issues in the kernel, the networking stack, alsa, cgroups, bluetooth, acpica, glibc/keyring, mesa/graphics drivers, etc. and things like that before they were able to be used by the general audience. i haven't had pulseaudio crash in years. --- and actually, now that i think about it, pulling from google/flickr wouldn't be more than 100 lines of code. use grilo to fetch flickr/google drive images together with the gnome-online-accounts credentials, and stick it in a gtkflowbox for presentation like all the other items.
|
|
#
?
Dec 3, 2013 06:39
|
|
|
Opinion Haver posted:also suspicious dish i see you a lot in grey forums and i think you're a good poster thanks! as i said before, my only goal is to get people to start thinking about these problems, and give me feedback. i don't expect gnome3 to be awesome for everybody out of the box. discussion is a lot more important when it's about real problems and thinking towards actual solutions, instead of lovely snark i read on hn four years ago. also, since i had to look it up, fyi everyone else "Opinion Haver" (great name btw) used to be "yaoi prophet". you're a good poster too.
|
|
#
?
Dec 3, 2013 06:44
|
|
|
Suspicious Dish posted:at the same time, visionary stuff like this is what got the desktop where it is. we had to deal with lots of low-level plumbing before so that steam could run itself on a solid platform without crashing: networkmanager, pulseaudio, systemd, bluez, upower, accountsservice, and dri/aiglx were all invented inside the gnome community, and they're the things that allow valve to consider building steamos. coincidentally, networkmanager, bluez 2.0, pulseaudio, and systemd are some of the most confusing and buggy parts of the platform bluez doesn't appear to be documented at all. error messages are cryptic when they're not outright meaningnless. troubleshooting process: write a python script that frobs dbus, fervent prayer systemd and *kit can die in a fire. that's not unix. i don't know what it is but it's not unix anymore Suspicious Dish posted:i haven't had pulseaudio crash in years. pulseaudio crashed twice on me today. i'm gonna guess version 4.0 isn't new enough and there's a 4.2.1.2 that i was supposed to be using. (that said, unlike the other things in that list, pulseaudio has concrete benefits for me personally so i will suffer through the crashing and mysterious audio problems and occasionally poor quality. basically it's like linux sound circa 1995 wooo)
|
|
#
?
Dec 3, 2013 06:49
|
|
|
Notorious b.s.d. posted:coincidentally, networkmanager, bluez 2.0, pulseaudio, and systemd are some of the most confusing and buggy parts of the platform i believe you mean bluez5 http://www.bluez.org/bluez-5-api-introduction-and-porting-guide/ http://lwn.net/Articles/531133/ it's a fairly standard interface. it's a dbus one using dbus conventions. can you give an example of a cryptic error message you encountered recently? our bluetooth guy recently did an investigation for a bunch of error messages like this and tried to fix them up. i'll double-check that he fixed the ones that you got. Notorious b.s.d. posted:systemd and *kit can die in a fire. that's not unix. i don't know what it is but it's not unix anymore yes, you are correct: it's not an operating system invented over 40 years ago. it is also not minix, tru64 or os x. why do you care? Notorious b.s.d. posted:pulseaudio crashed twice on me today. i'm gonna guess version 4.0 isn't new enough and there's a 4.2.1.2 that i was supposed to be using. so, i'm going to ask: did you file a bug, potentially using your distro's automatic crash handling tool? we investigate all crashes reported through that system, and reports for pulseaudio crashes for the last 3 months aren't very numerous. look at the main hot problems page to see the kind of data we get normally.
|
|
#
?
Dec 3, 2013 07:02
|
|
|
networkmanager is still fairly buggy, but a lot of this is growing pains due to networking just being actually loving complicated, and the fact that they exposed way too much of their api as public out of the gate, so the tech inside has been a bit stilted as a result. i'm constantly working with the networkmanager guys and reporting bugs since i maintain the network menu in the top bar of the shell. you'd be surprised how often hardware vendors just get poo poo horribly wrong and a surprising number of bugs just end up being stupid hardware they have to hack around in the kernel. also, we often get contracts to implement "the next big thing" in terms of networking protocols (wimax! infiniband! evolved hspa!) but they never get deployed widely. like, hardware never works properly but it seems networking companies just poo poo out hardware that is more broken than anything else on the planet.
|
|
#
?
Dec 3, 2013 07:05
|
|
|
Suspicious Dish posted:i believe you mean bluez5 no i am pretty sure i meant 2.0. or 3.0? the dbus stuff came in with a total rewrite that was a massive bundle of regressions for bluetooth on linux. my most recent bluez guru meditation was "input service timed out." i don't know what an input service is, or how i can monitor its status, or where it logs. and none of these things are google-able. i've been using and writing software on unix and linux systems since i was like 12, and i can't make bluetooth work reliably. what does that tell you about the usability story? Suspicious Dish posted:so, i'm going to ask: did you file a bug, potentially using your distro's automatic crash handling tool? we investigate all crashes reported through that system, and reports for pulseaudio crashes for the last 3 months aren't very numerous. look at the main hot problems page to see the kind of data we get normally. hell no i didn't the last time i reported a bug in linux desktop software, it was ignored for two years and fixed by accident. (that was a gnome 3 product btw) i only reported that one because it was a persistent thorn in my side. a buggy sack of poo poo that crashes on occasion isn't worth the hassle to me as an end-user, your odds of getting any benefit from bug reports are about zero. unless you can triage yourself and mail a patch, more often than not, nothing happens (if you do mail a patch, odds are pretty good you get yelled at or ignored)
|
|
#
?
Dec 3, 2013 07:08
|
|
|
also before you ask, yes that retrace app is super slow. it queries the red hat bugzilla api super stupidly (and red hat bugzilla is a really weird internal slow fork with some bizzaro stuff. our internal rhel processes are tied closely into custom bugzilla extensions). no, i dont work on "developer experience", the team that produces abrt and the retrace web app. yes, i filed a bug about it being slow.
|
|
#
?
Dec 3, 2013 07:10
|
|
|
i'm just gonna leave this here http://www.jwz.org/doc/cadt.html
|
|
#
?
Dec 3, 2013 07:11
|
|
|
Notorious b.s.d. posted:hell no i didn't i read every piece of bugmail i get, and try and debug problems after they're reported. i constantly check for crashers on the retrace app, and look for high-priority bugs in the rhbz and gnome bz. the desktop team at red hat has an internal weekly meeting where we discuss high-priority bugs, and crasher bugs are always top priority. reporting a crasher bug is a bit annoying because abrt is terrible right now, but we do investigate all abrt crashers we get. can you give any more details about how pulseaudio crashed? perhaps a backtrace? what distro are you on?
|
|
#
?
Dec 3, 2013 07:14
|
|
|
Suspicious Dish posted:and actually, now that i think about it, pulling from google/flickr wouldn't be more than 100 lines of code. use grilo to fetch flickr/google drive images together with the gnome-online-accounts credentials, and stick it in a gtkflowbox for presentation like all the other items. except gnome-online-accounts can't actually log in to google
|
|
#
?
Dec 3, 2013 07:15
|
|
|
Notorious b.s.d. posted:i'm just gonna leave this here yes, we've all read that one. i get around 10 new bugs a day on average. i try to personally debug all issues, but sometimes they're with obscure hardware i don't have access to (it only crashes with an amd chip, or with a soundblaster 3000 or something). we might hear about upstream fixes for something like this, or rework code to be more spec-compliant, and ask you to retest with a newer version as part of the debugging process. we rarely obsolete products just because we feel like rewriting them nowadays -- 90% of the tech introduced in gnome3 was the same as before, the big new component was gnome-shell instead of gnome-panel. next cycle we're planning on obsoleting gnome-menus because the menu-spec invented by kde is a giant piece of poo poo that does this weird rear end lisp-in-xml convoluted nonsense that makes it really hard to build an editor that lets users reorganize where their apps are. for that, we'll probably deprecate gnome-menus (along with alacarte, the ridiculously complicated aforementioned editor) and reassign maintainership to the people from mate/cinnamon if they want to reuse it. otherwise, the code will die.
|
|
#
?
Dec 3, 2013 07:19
|
|
|
pseudorandom name posted:except gnome-online-accounts can't actually log in to google   ???
|
|
#
?
Dec 3, 2013 07:21
|
|
|
Suspicious Dish posted:yes, we've all read that one. i get around 10 new bugs a day on average. i try to personally debug all issues, but sometimes they're with obscure hardware i don't have access to (it only crashes with an amd chip, or with a soundblaster 3000 or something). we might hear about upstream fixes for something like this, or rework code to be more spec-compliant, and ask you to retest with a newer version as part of the debugging process. how is deprecating a major part of the desktop not a rewrite for the sake of a rewrite? gnome 3 also introduced gtk 3, which was riddled with regressions. it may not be 90% of the tech by lines of code, but in terms of user impact it was a Very Big Deal
|
|
#
?
Dec 3, 2013 07:30
|
|
|
fundamentally open source is driven by a.) commercial interest and b.) scratching your own itch, even if it frustrates your users. open source desktop environments are the worst intersection of low commercial value and idiosyncratic project direction. i praised kde earlier, but kde 4 was the same kind of horrible regression filled disaster as gnome 3. they just recovered sooner. kde 4.10 actually reaches feature parity with 3.5. just in time for a new set of rewrites and regressions in kde 5 Notorious b.s.d. fucked around with this message at 07:38 on Dec 3, 2013 |
|
#
?
Dec 3, 2013 07:31
|
|
|
and then you enter your username and password and it immediately demands your username and password again, ad infinitum almost as if nobody involved tested it
|
|
#
?
Dec 3, 2013 07:33
|
|
|
Notorious b.s.d. posted:how is deprecating a major part of the desktop not a rewrite for the sake of a rewrite? user testing and ui feedback were showing that the panel wasn't meeting our ux goals. in fact, the gnome-shell codebase started with a prototype from the ux hackfest where we were able to give users something super early to play around with. http://www.vuntz.net/journal/post/2008/10/22/494-desktop-shell-from-the-user-experience-hackfest-general-overview http://blog.fishsoup.net/2008/11/22/gnome-shell-status/ it evolved organically over time, adapting to feedback from user testing research and gnome contributors. Notorious b.s.d. posted:gnome 3 also introduced gtk 3, which was riddled with regressions. it may not be 90% of the tech by lines of code, but in terms of user impact it was a Very Big Deal gtk3 removed a bunch of api deprecated since gtk 2.8, that was unmaintained and underperforming. we also rewrote the custom theme engine system and an ad-hoc format with standard css (e.g. border-radius: 5px; Just Works). those are the two major changes in gtk3.
|
|
#
?
Dec 3, 2013 07:40
|
|
|
wait which one of you works for rhel again.
|
|
#
?
Dec 3, 2013 07:41
|
|
|
pseudorandom name posted:and then you enter your username and password and it immediately demands your username and password again, ad infinitum i just tested it now and it worked fine. there was a bug on google's end a little while ago where they disabled one of their documented oauth2 login types without any warning. we switched to another, but they reenabled it since. if you just tested it and it doesn't work, please tell me what version you're on (3.8.4?) so i can try it out locally and see if it's broken here.
|
|
#
?
Dec 3, 2013 07:42
|
|
|
FamDav posted:wait which one of you works for rhel again. i work for red hat. feel free to complain about rhel all you want at me.
|
|
#
?
Dec 3, 2013 07:42
|
|
|
Suspicious Dish posted:user testing and ui feedback were showing that the panel wasn't meeting our ux goals. in fact, the gnome-shell codebase started with a prototype from the ux hackfest where we were able to give users something super early to play around with. just to hang a hat on it: ux goals are more important than NOT breaking a standard, because the standard is ugly. this is a short explanation of why gnome 3 is like it is. Suspicious Dish posted:gtk3 removed a bunch of api deprecated since gtk 2.8, that was unmaintained and underperforming. we also rewrote the custom theme engine system and an ad-hoc format with standard css (e.g. border-radius: 5px; Just Works). those are the two major changes in gtk3. this may have been the DESIRED changes, but there was a lot more breakage than theme engines
|
|
#
?
Dec 3, 2013 07:43
|
|
|
Suspicious Dish posted:i work for red hat. feel free to complain about rhel all you want at me. rhel still owns even if gnome 3 is terrible
|
|
#
?
Dec 3, 2013 07:44
|
|
|
Suspicious Dish posted:i work for red hat. feel free to complain about rhel all you want at me. i fuckin love rhel you get a lot of bitches in westford tho?
|
|
#
?
Dec 3, 2013 07:45
|
|
|
FamDav posted:you get a lot of bitches in westford tho? i'm moving into my new apartment in cambridge tomorrow
|
|
#
?
Dec 3, 2013 07:48
|
|
|
Notorious b.s.d. posted:just to hang a hat on it: ux goals are more important than NOT breaking a standard, because the standard is ugly. this is a short explanation of why gnome 3 is like it is. making sure that people can use our software is actually important, yes.
|
|
#
?
Dec 3, 2013 07:48
|
|
|
Suspicious Dish posted:i just tested it now and it worked fine. https://bugzilla.redhat.com/show_bug.cgi?id=888822
|
|
#
?
Dec 3, 2013 07:49
|
|
|
ah, 2FA. that was my other guess. basically, google didnt support it for caldav for a long time, so we couldnt actually add it to the framework. 2FA for caldav finally got supported back in mid-september. we're in the middle of backporting everything into f19, and the next compose is this week, so hopefully the fix will make it.
|
|
#
?
Dec 3, 2013 07:52
|
|
|
|
| # ? Jun 1, 2024 07:58 |
|
|
btw mr. bsd i'm still waiting your reply about systemd and how "it's not unix" and why that matters and for details about your supposed pulseaudio crashes. if they happened today you should still have the core dumps around and be able to paste a backtrace.
|
|
#
?
Dec 3, 2013 07:53
|
|