|
Holy poo poo that's amazing, now I want to do work poo poo over the weekend
|
# ? Jul 29, 2017 06:55 |
|
|
# ? Apr 28, 2024 20:40 |
|
cinci zoo sniper posted:i had some unironic gentoo fanatic try to cry at me how systemd is terrible garbage and how sysvinit, openrc, and upstart are all undeniably better than it gentoo with systemd is fantastic, highly recommend get that openrc trash out of profile/packages
|
# ? Jul 29, 2017 08:54 |
|
Tankakern posted:gentoo ... is fantastic lol no
|
# ? Jul 29, 2017 18:17 |
|
from what I remember about working with gentoo in rhe distant past, it's pretty cool and good for learning how to build your own distro (sort of) and really get in the muck with how things work but now I have other stuff to do with my life so I'll stick with redhat and debian, thanks
|
# ? Jul 29, 2017 19:16 |
|
code:
|
# ? Jul 29, 2017 19:29 |
|
what is that
|
# ? Jul 29, 2017 20:51 |
|
probably system build templates for gentoo or something does it still take literally a week for gentoo to build and install?
|
# ? Jul 29, 2017 21:15 |
|
it's a list over what system profiles you have available on x86-64 on gentoo. my point was showing what's available, and to show that systemd's there, on several profiles. and no, we're not in the age of pentium III anymore, so it does not take literally a week. you can think on it like.. a full blown kde install with > 1000 packages might take about 24 hours. the packages ruining the compile time stats are: libreoffice, chromium, and anything that's bundling/building webkit. chromium is by far the worst, there's like over 27000 files to compile (not an exaggeration) before you get your shiny web browser.
|
# ? Jul 29, 2017 22:01 |
|
I remember running make world on a 486 running freebsd at a time when it was popular for people to run 486es as routers to share connections it was still building a week later
|
# ? Jul 29, 2017 22:04 |
|
how long for just chromium?
|
# ? Jul 29, 2017 22:05 |
|
Truga posted:probably system build templates for gentoo or something thanks! Tankakern posted:it's a list over what system profiles you have available on x86-64 on gentoo. oh
|
# ? Jul 29, 2017 22:05 |
|
every time i want to see what's new in chromium, i leave it overnight if i'm not on a skylake or something. trying it on a carrizo amd apu now, think it's been over 6 hours and it's still not done
|
# ? Jul 29, 2017 22:10 |
|
why wouldn't you just download a binary like a normal human being what does compiling it yourself get you other than hastening global warming
|
# ? Jul 29, 2017 22:59 |
|
Suspicious Dish posted:This Issue from the specification for GLX_EXT_texture_from_pixmap offers us a pretty big clue: let me see if i'm picking this up correctly in a composite-enabled x11 server with the nvidia driver, the "off-screen" storage for composite's "window hierarchies" is gonna be textures so every time the screen is drawn on at all, it's gonna use the glx tfp extension to turn those textures into a drawable and immediately render it to the screen... and that stuff can never be read back. it's a write only medium bear in mind i am not a graphics dude, i'm just curious about who broke what and why
|
# ? Jul 29, 2017 23:15 |
|
BobHoward posted:why wouldn't you just download a binary like a normal human being if you have your compile flags set up right, then the system will run faster than generic pre-compiled binaries not enough faster to ever make up for the compile time you have to invest, but faster https://fun.irq.dk/funroll-loops.org/
|
# ? Jul 29, 2017 23:59 |
|
RFC2324 posted:if you have your compile flags set up right this is the impossible condition it's not like linux distributors choose the wrong flags on purpose. often the flags they chose are already the best choice, and you will eke out nothing more. other times you will manage to get some subset of the system to compile with -Os -fomit-frame-pointer --gently caress-safety and it works except for a feature you didn't test
|
# ? Jul 30, 2017 00:03 |
|
BobHoward posted:why wouldn't you just download a binary like a normal human being it gets you absolutely nothing useful some people jack off to the idea that they can disable PAM system-wide by changing a config file and re-compiling the entire system
|
# ? Jul 30, 2017 00:05 |
|
RFC2324 posted:if you have your compile flags set up right, then the system will run faster than generic pre-compiled binaries compile while you sleep
|
# ? Jul 30, 2017 00:09 |
|
Notorious b.s.d. posted:this is the impossible condition the distros compile generic flags that should be universally compatible, but if *I* set the flags that are best for *my* specific machine it will run much faster!!! the reality is that any gains you get are so marginal that you will never recover the time lost to compiling, and the only use gentoo has is teaching you to copy arcane commands from a wiki, and how to read logs so you can get directed to the right wiki page
|
# ? Jul 30, 2017 00:10 |
|
Notorious b.s.d. posted:let me see if i'm picking this up correctly There's off-screen storage, but it's managed through the magic of the NVIDIA libGL.so (and kernel, I imagine) driver. Usually, the flow looks like this: 1. XCompositeRedirectWindow (turn it off screen) 2. XCompositeNameWindowPixmap (get the backing pixmap for the window) 3. User sends SIGSTOP Now, once per frame: 4. glXBindTexImageEXT to turn the backing pixmap into a GL texture 5. glDrawElements to texture the texture 6. glXSwapBuffers Assuming that the client rendering isn't using X11 to draw at all (e.g. override-redirect glxgears) I'm saying that this will continue to work. glxgears renders into an off-screen renderable, unclipped texture, and the CM can rebind and redraw the updated window contents without needing the X server's involvement at all. This also assumes we are rebinding every frame and not using Damage to track when to rebind a window. xcompmgr is stupid and doesn't use Damage IIRC.
|
# ? Jul 30, 2017 06:15 |
|
RFC2324 posted:if you have your compile flags set up right, then the system will run faster than generic pre-compiled binaries i take issue with the idea that the bolded is even true in the first place
|
# ? Jul 30, 2017 06:24 |
|
BobHoward posted:i take issue with the idea that the bolded is even true in the first place well, if the flags were wrong to start with, sure the system will run faster. for example it is usually a bad idea to tell gcc to emit code optimized for an i386. good thing nobody is doing that on their binary packages.
|
# ? Jul 30, 2017 06:33 |
|
it's not just about the processor-specific optimizations, it's also that you're given the tools to micromanage your system, by the magic of use flags. gentoo is a kind of meta-distribution, you can mend it to any form you like. but you have to like doing such mending to get any out of it.
|
# ? Jul 30, 2017 10:34 |
https://ring.cx/en/news ??
|
|
# ? Jul 30, 2017 12:11 |
|
wish people would stop addressing stallman as dr (or at all)
|
# ? Jul 30, 2017 12:29 |
Gazpacho posted:wish people would stop addressing stallman as dr (or at all) gnu dr.
|
|
# ? Jul 30, 2017 12:31 |
|
red hat just signed the death warrant for butterfsquote:The Btrfs file system has been in Technology Preview state since the initial release of Red Hat Enterprise Linux 6. Red Hat will not be moving Btrfs to a fully supported feature and it will be removed in a future major release of Red Hat Enterprise Linux.
|
# ? Aug 1, 2017 19:25 |
|
good, brtfs sucks sorry lennart
|
# ? Aug 1, 2017 19:35 |
|
cool linux stuck on ext until the heat death of the universe
|
# ? Aug 1, 2017 19:47 |
|
pram posted:cool linux stuck on ext until the heat death of the universe Redhat defaults to XFS
|
# ? Aug 1, 2017 19:50 |
|
when i think cutting edge i definitely think of a 20 year old filesystem made by a defunct corporation that cant take snapshots
|
# ? Aug 1, 2017 19:53 |
|
hfs made it through 32 years. one of those areas where pragmatism certainly rules, don't mess with stuff that works without very good reason, and btrfs just needs the murder to fulfill the reiser4 pattern of simply never stabilizing quicker than it is further complicated
|
# ? Aug 1, 2017 20:10 |
|
yes but now osx will have apfs which is the worlds most advanced desktop filesystem
|
# ? Aug 1, 2017 20:12 |
|
pram posted:yes but now osx will have apfs which is the worlds most advanced desktop filesystem unironically true
|
# ? Aug 1, 2017 20:51 |
|
pram posted:cool linux stuck on ext until the heat death of the universe Probably launching their own new FS, especially after recently buying Permabit.
|
# ? Aug 1, 2017 22:30 |
|
Found out that a service cannot talk to mysql, so I shelled into it to debug it running ubuntu 16.04.1, with the 4.4.0-67 kernel apt-get is complaining that it cannot rename files because there is no free space the target directory is the root device with 27G total space, 11G used, 17G free the root user can create a 64MB file on that device with no issues what's the issue? btrfs I thought I changed it to ext4 but I guess I gave it a second chance. Not anymore.
|
# ? Aug 2, 2017 04:33 |
|
a systemctl reboot fixed it it's gonna get wiped and replaced with ext4 though
|
# ? Aug 2, 2017 04:38 |
|
maybe ou 're running out of inodes
|
# ? Aug 2, 2017 04:42 |
|
pram posted:when i think cutting edge i definitely think of a 20 year old filesystem made by a defunct corporation that cant take snapshots xfs is actively maintained snapshottimg is provided by lvm, not the file system driver
|
# ? Aug 2, 2017 04:47 |
|
|
# ? Apr 28, 2024 20:40 |
|
btrfs might be fine technically or whatever but it's been in the kernel for 8 years and raid5 isn't finished. zfs has their own reasons but linus should have told the btrfs crew to put a sock in it and come back with something worthwhile
|
# ? Aug 2, 2017 04:47 |