IRC Logs for #crux Thursday, 2015-01-22

korigrr openmailbox is down00:00
diversedoing another maintenance?00:00
koridiverse: if "Error: Unable to connect to MySQL and select database. MySQL reported: Can't connect to MySQL server on 'openmailbox-sql1' (111 "Connection refused")." is maintenance, sure00:01
diversewell the last time they took it down was for upgrading00:02
*** BitPuffin has quit IRC00:03
koriwell, at least pierre acts fast00:06
koriit's back up00:06
*** hhhhhhhh_ has quit IRC00:10
diversethat's didn't take long00:15
diverse*that00:15
*** EmoSpice_Home has joined #crux00:18
diversewell I solved my C binding linker issue thanks to hhhhhhhh00:29
diversenow it's time to uncompress shit00:29
*** EmoSpice_Home has quit IRC00:54
*** tilman has quit IRC01:03
*** tilman has joined #crux01:05
*** jue_ has joined #crux01:55
*** Workster_ has joined #crux01:55
*** Workster has quit IRC01:55
*** Workster has joined #crux01:55
*** fnords has joined #crux01:57
*** RobinStamer has joined #crux01:57
*** rauz_ has joined #crux01:58
*** Feigrim_ has joined #crux01:59
*** nthwyatt has joined #crux01:59
*** nogagplz_ has joined #crux01:59
*** heroux_ has joined #crux01:59
*** rmull_ has joined #crux02:00
diverseoh wow, 2 more users and we hit 90! \o/02:01
*** tvaalen_ has joined #crux02:02
*** kInOzAwA has quit IRC02:03
*** kInOzAwA_ has joined #crux02:03
*** Feigrim has quit IRC02:06
*** jue has quit IRC02:06
*** darfo has quit IRC02:06
*** z3bra has joined #crux02:08
koridiverse: well, i don't think we're hitting 90 until they come back, haha02:08
diverseheh02:08
diverseRomster: stuff now decompresses, rebase02:08
diverseand also install flate2-rs from my repo02:09
diverseso with that and some bug smashing, I should have a fully functional pkgadd, pkgrm, pkginfo alternative02:11
*** jdolan has quit IRC02:14
*** jdolan has joined #crux02:14
diversekori: btw, what's your impression on neovim so far?02:19
korinot much different than regular vim for me, other than the fact that it feels snappier02:19
diversecan't wait02:19
koriwait for what?02:20
diverseI mean, looking forward to it02:21
koriif you use vim currently, there's not much cost to move to  neovim.02:21
korimight as well do it now02:22
diverseI rely on gvim heavily as well, unfortunately.02:22
diverseunless you know of a third party gui plugin that someone made?02:22
korinope02:22
b1n0will try neovim when I finish "basic" stuff here02:31
diverseI do see a lot of potential for neovim, but I guess the question for me would be, can I live without the gui part (at least for now)02:39
koridiverse: what do you require that is gui exclusive02:40
diversekori: I like to have more colors than 256, and the I use the cursorline feature and because my cursor is shaped like an underscore, (I hate the block cursor covering whole characters when editing), cursorline looks just like one long "__________" under the characters02:43
koridiverse: termite supports that underline02:44
korithing02:44
korihttps://u.teknik.io/y5Hl2g.png02:44
rmullI looked at the page at https://github.com/neovim/neovim/wiki/Differences-from-vim but there is nothing that seems very compelling02:45
rmullAm I missing something?02:45
diversekori: yeah I have that in lxterminal, but that's not the problem02:45
korirmull: the planned features, aye02:45
rmullhmm02:46
diversermull: it has a flexible architecture for guis and does operations (like mode switching) better02:46
rmullI will keep it on my radar02:46
diverseit does have a lot of promise02:46
rmullcool02:47
rmullI guess I've always just taken vim for granted02:47
diversevim is great, it's just that this appears to be better ;)02:47
rmullI see that they removed some legacy features - could one say that mplayer is to mpv as vim is to neovim?02:47
diverseyeah, you can think of it like that02:48
rmullNice02:48
rmullSolving problems I didn't know I had ^_^02:48
diversebecause I imagine it's also taking in a lot of the patches that never get into vim, like multithreading support02:48
diversealso it's kind of funny how Bram thought refactoring the project to be a was "not the solution" and that you should "improve what we already have" since he never really bothers to accept many patches to begin with :P02:49
rmullThe community will make it's choice02:49
rmullits.02:50
diverseyeah02:50
diverseand there is a lot of things vim can do already like "set nocompatible" by default02:53
koriI enjoy clean up efforts02:54
koribut what I enjoy more is programs that are clean from the start02:54
diverseyeah definitely, who still needs support for really old OSes?02:55
diversethat just burdens the project with more maintenance02:55
diverseand possible indirect bugs02:56
diversermull: so yeah that's a good way to view it, it's like mpv02:57
rmullanybody getting a port going?02:58
diversekori has one02:58
koriall the dependencies also :D02:58
korinote: you need to update msgpack-c before you update neovim, every time02:58
diversemsgpack-c?02:59
rmullkori, are you not on portdb?02:59
korirmull: I was, but then I dropped httpup support for my repo02:59
koribecause I couldn't be arsed to update REPO every time02:59
rmullhaha03:01
rmullgit hook?03:02
koriughhh03:02
korithe git driver is faster anyway03:02
rmullcan crux be set up to sync all its ports over git?03:03
rmullI should do that03:03
diversethat's the point of the git driver kori worked on03:03
korirmull: ohhh03:03
korinow that's something I didn't think of03:04
koriyes, probably03:04
koriactually03:04
koriI don't see why you wouldn't be able to03:04
rmullwhere can I find the git driver?03:05
diversermull: I guess you could help him out with that? :)03:05
korihttps://github.com/6c37/cross/blob/master/tools/git03:05
rmulldiverse: Shouldn't it be just as simple as putting the right file in /etc/ports?03:05
korihttps://github.com/6c37/cross/blob/master/tools/6c37.git sync file for our repo03:05
rmullkori: Thansk03:06
koripoke around! we have some High Quality™ ports03:06
diversekori: you should put that as the repo description ;)03:07
korihttps://u.teknik.io/GCX7TY.png03:09
korimay be too much03:09
korihah03:09
koriI'll just keep "Our CRUX ports."03:10
rmullHow does the git driver know where to check the ports out?03:12
diversejust say "High Quality™ CRUX ports" if you say "Our" you make your repo feel exclusive instead of inclusive03:12
rmullWhat I mean is - is there any way to specify a destination?03:12
rmullOr should I add that03:13
korirmull: it always is $PORTS_DIR/$NAME03:13
kori$PORTS_DIR being defined in ports(8)03:13
rmullhmmm03:14
rmulloh there we go. Having an empty /usr/ports/core causes it to fail03:14
rmullbecause it's not a git repo I guess03:15
*** Feigrim_ has quit IRC03:15
rmullGonna switch my whole system to git :D03:15
diverse:)03:16
koriwhat a wonderful time to be a live03:17
koriI love CRUX so much03:17
rmullThis seems like it'd be a decent wiki topic03:18
rmullports over git I mean03:18
diversegoing to write the procedures on the wiki?03:19
koriI'd like to just take a day and go ham on the wiki03:19
rmullSure, unless someone else wants to03:19
rmullmaybe git requires more system resources on the server and that if everyone switched to it it'd be a lot of IO and CPU03:22
rmullshrug, whatever03:22
diversewell it's good to have a discussion first03:23
diversegit should be pretty fast in general if has been through a lot of optimization and testing ever since it was made03:24
rmullhm, something is weird03:27
rmullWhat's with the git reset --hard?03:27
rmullWouldn't that purge all the downloaded and built files?03:28
diverseah that, that's just a permanent way to undo what you did depend how far you want to go back in the commit history.03:28
rmullThat's no good for general purpose use, though03:28
rmullAlso, the branches crux uses aren't just master03:28
rmullso when I ports -u, it deletes my built stuff and hides all the actual ports and shows just a README, lol03:29
rmullI'm gonna remove the reset03:29
diversewhere is "git reset --hard" written?03:29
diverse"reset" is the more dangerous way, "revert" is the safest03:29
korirmull: you shouldn't keep custom ports in the working tree anyway.03:29
korias in03:29
koriwell03:29
koriin the regular sync tree03:29
korigrr you get what I mean03:30
rmullIn the git driver03:30
*** mavrick61 has quit IRC03:30
diverseeww... that would probably be undesirable...03:30
diverserevert might be better03:30
rmullActually, even the git fetch doesn't work for crux ports03:30
*** mavrick61 has joined #crux03:31
koriI've actually discussed this03:31
diversewith who?03:31
korijaeger (I believe)03:31
diverseoh I see now03:32
diverseso it's like ports -u where it unsets any user made changes03:33
koridiverse: yes03:33
koriI just copied the behavior and kept it because it nullifies breakage due to mismatches between HEAD and the current tree03:34
koriI just recommend keeping a custom port dir03:34
korias an overlaty03:34
korioverlay'03:34
diverseyeah03:34
koriif you must make any edits.03:34
rmullIn the case of the crux ports, the git fetch isn't enough to bring up the ports tree03:35
rmullI needed to first do a git pull03:35
rmullpull is just a fetch and merge I think, so I think the missing step was getting onto the right branch03:35
diversewell actually what you want to still have is, say for ports for 2bwm or st where they have a config.h file, let the user still have the backup copy if they made one03:36
rmullor you could just keep a local git branch that modifies the crux ports when merged to the crux branch03:36
korirmull: well, if you're first bringing up the ports, it does a git clone03:36
korias in, you wouldn't have /usr/ports/core/03:36
rmullright03:36
diverse--hard03:38
diverse Resets the index and working tree. Any changes to tracked files03:38
diverse in the working tree since <commit> are discarded.03:38
diverseokay, that sounds fine to me03:39
rmullWhy is that fine?03:39
diversejust the tracked files03:39
rmulloh oh oh03:39
rmullRight.03:39
rmullsorry03:39
diversebackups copies shouldn't get removed03:39
korirmull: I believe that to be fine since you should make a custom ports overlay anyway03:39
rmullso untracked dowloads and built files are untouched03:39
koriit's probably the best practice03:40
diversermull: that's what I read it as03:40
rmullkori: With git, I don't think the ports overlay is necessary03:40
korirmull: I still keep one03:40
rmullI'll just keep my changes committed to my own local branch03:40
korigit is confusing, man03:41
diversegit grows on you the more you use it ;)03:41
rmullI taught a git class once03:41
rmull(for beginners)03:42
rmullI have never participated in a large scale distributed development effort though, so I have no practice03:42
diverseme neither but every since I started my project, I learned more about git03:42
diverse*ever03:42
diverseand I find other neat things about it03:43
rmullUgh, no, reset --hard is no good for the crux ports03:43
rmullit reverts the to README-only03:43
diversehuh?03:44
rmullThat confuses me a bit03:44
korias I understand it, reset --hard removes all customization03:44
diverseof the tracked file03:44
koriof tracked files03:44
koriyes.03:44
rmullThat's not only it03:44
rmullYou should just try it with one of the crux ports repos03:45
rmullIt'll clone okay03:45
rmullbut when issuing ports -u and it runs the fetch and reset, the filesystem contents all get hidden and the only thing showing is a README03:45
rmullbranch -a shows that I'm on a local branch called 3.103:45
diverseI wonder what the .gitignore says?03:46
rmulland that I'm behind origin/3.1 by 9424 commits (for opt)03:46
rmullIf I do a git pull, the filesystem contents come back03:46
korinow if you gentlemen excuse me, I'm going to do a make mrproper kernel03:47
korifor learning purposes03:47
rmullbranch -a still shows I'm on 3.1 local branch03:47
*** jdolan has quit IRC03:47
mikelkori: so starting from a blank .config?03:48
korimikel: not exactly blank03:48
korithe kernel has some vital elements enabled03:48
diverseI think that resets it to the defaults03:48
mikelmmm03:48
rmullif I remove the reset --hard, everything is okay03:48
rmulldiverse: I don't see a .gitignore actually03:48
rmullYeah, but there's something weird about the branches that is getting it screwed up03:48
rmullI'm cool running without the reset --hard myself, I'm just going to use it without that03:48
mikeldefaults tend to work fine for me barring a few expansion card drivers03:48
rmullI encourage you to try it though, maybe I'm missing something03:49
mikelexcept forgetting to install modules...03:49
*** jdolan has joined #crux03:49
rmullI won't be adding anything to the wiki until it's settled03:49
rmullGotta run, catch you all later03:49
korirmull: that's for the best03:49
korithe git driver requires testing, anyway03:49
kori:DDDD03:49
diversehmm, how about instead of a git fetch, how about a git pull?03:55
diversemaybe what it needs is really a pull and then reset03:56
koridiverse: used to do that, jaeger recommended git fetch03:59
diversekori: also for debugging purposes, there is no need for a -q (--quiet) option to be there until it is ready for release, if anything, put a -v there, so it shows more of what is going on.04:00
diversethat way it can help rmull out ;)04:03
koribrb, reboot!!04:03
*** kori has quit IRC04:04
*** Feigrim has joined #crux04:06
diverseFeigrim: how is it going?04:07
*** kori has joined #crux04:34
*** kori has quit IRC04:34
*** kori has joined #crux04:34
koritook a couple tries to get the ethernet driver right :P04:34
koriprecisely, 804:34
korigrr gotta get my mouse figured out now04:35
korihere's to 16 tries04:35
diversehuh, shouldn't you have the evdev module enabled?04:36
koriEVDEV04:36
koriI forgot to check that.04:36
koriit's on04:36
diversehuh I thought the defaults would have that enabled by default04:36
koriweird.04:36
diversekori: could be a hardware thing, try plugging the mouse into a different usb port04:39
koridiverse: nay04:39
koriit's working fine04:39
koriI believe I forgot to enable something04:39
diversewell I thought I would add my thoughts to troubleshooting, but whatever04:40
korioh I didn't mean to sound rude, sorry04:42
koribut yes, I did test a different port, before you mentioned it04:42
koriI should have mentioned that04:42
kori:D04:42
*** kori has quit IRC04:46
*** Feksclaus has quit IRC04:48
*** kori has joined #crux04:48
*** kori has quit IRC04:48
*** kori has joined #crux04:48
korithis might be a xorg issue.04:49
*** Feksclaus has joined #crux04:51
diversetry rebuilding the evdev port and restart the X session04:54
diverse(if you haven't already)04:54
koriI actually tried that04:55
koriI'm confused, really04:55
diversewell I did have your problem before, and it was solved from unplugging from a usb2 port to a usb3 one... don't ask04:56
koriuh04:56
korihold on.04:56
koriTHAT WORKED04:57
koriLMAO04:57
diverseand yet here comes yet another one of kori's mysterious revolutions(TM)04:57
kori truly amazing04:57
korinow I have this very slim kernel that should get slimmer04:57
diverseso what solved it?04:58
koridoing what you did04:58
korilol04:58
diversethe usb3 thing?04:58
koriyes04:58
korihaha04:58
koriI did NOT expect that to work04:58
koriwell that was fun. thanks diverse04:59
diverseglad I can help04:59
diversekori: just a query, what motherboard and psu are you using?05:02
koriI actually forgot and I forgot how to find that out05:03
diversealright, do you know how many watts your psu uses?05:04
diverse(just simply open the cover of your case and look at the psu if you don't know)05:06
korithe way my desk is set up doesn't facilitate me opening my case right now05:08
korinext time I shut down, I'll let you know!05:08
diversealrighty, I guess I'll start adding help usages to my commands05:19
*** SiFuh has quit IRC05:52
*** dougiel has quit IRC06:10
*** phant0mas has joined #crux06:56
Romsteri'm tired07:27
*** Feksclaus has quit IRC07:29
diverseRomster: rough day at work?07:29
Romsternot really just haven't been sleeping well has accumulated07:30
diverseinsomnia?07:31
Romsterlate nights and its been warmer07:31
*** leo-unglaub has joined #crux07:31
diverseI guess you got an opportunity to get some more sleep today :)07:32
Romstermaybe it is cooler now07:32
diverseturn on the AC too?07:33
diverseis it summer over there?07:34
Romsteryeah07:34
leo-unglaubhey diverse :)07:35
diversehey leo-unglaub07:35
leo-unglaubdiverse: so much for the awesome polarssl *g* https://polarssl.org/tech-updates/security-advisories/polarssl-security-advisory-2014-0407:35
diversenothing can be perfect, but I'm sure polarssl is still the most secure07:36
leo-unglaubnow you can say: That would not have happened with rust07:36
leo-unglaub;)07:36
diversewas it a memory exploit?07:36
leo-unglaubyep ;)07:36
leo-unglaubmemset( cur->next, 0, sizeof( asn1_sequence ) );07:37
leo-unglaubthat was missing07:37
diverseah07:37
Romsterif only the compiler would error.07:37
Romsterthey just cloned openssl no and then started rewriting bits?07:38
leo-unglaubRomster: not libressl07:39
leo-unglaubi am talking about polarssl07:39
Romsteroh07:39
Romsterclearly tired07:39
diverselibressl isn't rewriting anything, they are creating another lib with a better API called libtls, which obviously uses the better protocol07:40
Romsteryeah someone needs to write it in rust.07:40
leo-unglaubRomster: hehe, go for it07:41
leo-unglaubi just have reimplemented SHA in Rust07:41
diversewell not just that but implement a better curl that uses either polarssl or libtls07:41
leo-unglaubdiverse: you can compile curl with polarssl07:42
diverseah right, but curl still sucks -_-07:42
leo-unglaubdiverse: ./configure --without-ssl --with-polarssl07:42
leo-unglaubbut yeah, the architecture of curl is howwible07:42
leo-unglaubhowever .. rewriting crypto in rust is hard07:43
diverseI want to punch it in the face07:43
leo-unglaubi would never dare to write crypto myself07:43
leo-unglauball those padding stuff07:43
leo-unglaubis so fucking hard to get right07:43
leo-unglaub...07:43
Romsteryeah i would never attempt i07:43
Romstert07:43
diverseit's hard, but there are some crypto guys in #rust talking about writing cryptography related code07:44
diverses/are/were/07:44
Romsterneed to07:44
Romsterhurry up ebay so i can go cook dinner.07:44
diverseheck even someone was trying to reimplement zlib in pure rust although compressions are a weaker form of crypto07:44
Romsterat least it has to decompress and match sums07:45
diversewell for now, I'm happy with flate2-rs07:45
diverseand I can use it for my build command07:46
Romsternow if it can work like pigz does07:46
Romstersuper fast compressing with 4 cores07:46
Romsterthan 107:46
diverseand at least I finally figured out how to work with C bindings07:48
Romsternice07:48
Romsterso libarchive <<07:48
diversethanks to hhhhhhhh07:48
diverselibarchive is not needed anyway07:48
Romsterwait until you work on pkg build07:49
diverses/pkg/mport/07:49
Romster-_-07:49
Romsteryou changed the binary name?07:49
diverseyes07:49
diverseI even said so in #crux-devel07:49
Romsterforgot07:49
Romsterthe amount of things i do07:50
diverseit's already, I understand you are very tired :)07:50
diverses/already/alright/07:50
diverseman, my typos have gone up :P07:50
Romsteraround me i am not surprised.07:51
diverseif IRC only had a way to edit previous messages, that would be awesome07:51
Romsteri did think of a way on a improved irc protocol.07:51
Romsterbut i'll never write one.07:51
diverseit's the one thing I like about Skype, it lets you go back to and edit your previous messages07:52
diverseI hope Tox implements that feature too07:53
Romsterah ya fucking .... meh hate bids on ebay07:55
diverseleo-unglaub: so what have you been doing lately?07:55
diverseI guess you implemented SHA, how was that?07:56
leo-unglaubdiverse: it was ... harder than i thought07:57
leo-unglaubtook me nearly 5 hours ...07:57
diverseand you completed it?07:57
leo-unglaubyep07:57
diversewow that's awesome07:57
leo-unglaubi hashed 10 000 random strings with another sha tool and compared it07:58
leo-unglaubit seams to be working07:58
diverseif you can host your code on github, I can probably find ways to write it cleaner08:02
diversethe existing the code08:02
diversealthough I guess you're a bitbucket fan :)08:03
diverseleo-unglaub: ^08:04
leo-unglaubyes, i am ... github is way to fancy for me. they always change the ui and i never find stuff08:05
leo-unglaubbitbucket is more robust :)08:05
diversebtw, thanks for the star08:06
leo-unglaubsure08:07
Romstersha1 or did you goto sha51208:09
leo-unglaubsha1 is for pussies ;)08:10
*** mhe has joined #crux08:10
diversehahaha08:10
Romsternyan08:11
diverseit's interesting that ARM owns polarssl now08:13
leo-unglaubdiverse: yep .. i did know it 3 month before the anouncement *g*08:14
leo-unglaubhe made a mistake *g* i wrote him a bug report via email and he replied from an @arm.com address *g*08:14
diversehehehe08:15
diversethat was the give away huh?08:15
leo-unglaubyep :)08:15
leo-unglaubnormally he replies from @offsparc08:16
leo-unglauband then he did reply from @arm.com *g*08:16
diversecan I see your code?08:20
diverseleo-unglaub: ^08:25
Romsterneed sauce08:25
leo-unglaubdiverse: nah, better not08:32
leo-unglaubit's not optimal08:32
leo-unglaubRomster: prt-get depinst sauce08:32
leo-unglaub;)08:32
diverse(and one day, hopefully, mport install --deps sauce)08:34
Romsterfinally bought a new mouse. waits for it next week08:43
diverseRomster: so what I do during the install is unpack it all (the only way) and iterate through the fileset to see if any reject files are not needed and remove the files are rejected... yeah I should unpack it first to some /tmp/mport_install dir and when done move it to /08:43
RomsterRoccat Kone XTD Optical Gaming Mouse08:43
Romsteryou should be able to list files with tar-rs08:45
diverseI do that at the Pkg::open() stage and store it in the object for later referencing08:45
Romsterbtw rejected files for merge go in /var/lib/pkg/rejected/08:46
Romsterfor rejmerge to take care of08:46
diverseyeah I did it set it up for that08:48
Romstertar -ztvf file.tar.gz <- list the files of tar.gz08:48
Romsterthe rust libs need to support this08:49
leo-unglaubRomster: or bsdtar -l -f <filename>08:49
diverseright, that (hopefully) goes into the FileSet during Pkg::open()08:49
Romsteror preferable bsdtar. but diverse is using tar-rs a rust rewrite of tar08:50
Romsteri haven't looked at the new dep yet for gz08:50
diverse(there are no bindings to libarchive anyway!)08:51
Romsteryet08:52
diversebut it doesn't matter anymore, libarchive extracts each file during iteration and decides to remove anyway. I just unpack it and sort through the logic from Pkg.info.files and move the files directly08:53
leo-unglaubsoo, i am alone in the office for now :)08:53
leo-unglaubtime to implement sha3 (Keccak)08:53
Romsternude streak down the office.08:53
Romsterwtf uses sha3?08:53
leo-unglaubno idea, but it sounds interresting08:55
diverseI guess in a security pov like leo, the question is, what doesn't use sha3 and how can I implement it for it?08:55
leo-unglaubdiverse: sha3 has a problem08:55
leo-unglaubsoftware implementations are slow08:55
diverseI bet08:56
leo-unglaubthats why almost everyone uses sha208:56
diversebecause it's good enough08:56
diverseuntil faster computers come along08:56
leo-unglaubwell, it's better than md5 *g*08:56
diverseheh08:56
leo-unglaubsha3 is fast, if it's hardware implemented08:57
diverseleo-unglaub: is that a poke to pkgutils? ;)08:57
leo-unglaubdiverse: nope, a poke to every php application i have seen in the last 5 years *g*08:57
leo-unglaubor HTTP AUTH *g*08:58
diverseouch08:58
diverseleo-unglaub: so for file checksum purposes, is sha1 fine?08:59
leo-unglaubdiverse: no, sha209:00
diverseleo-unglaub: what about for people who (still) use really slow computers?09:00
leo-unglaubdiverse: also not a problem09:01
leo-unglaubsha2 is possible on every old computer09:01
leo-unglaubif they can unpack an archive, they can do sha209:01
diversesounds good09:01
leo-unglaub"slow" with hashes means that you can only 300 hashes per secound instead of 500 hashes per secound09:01
leo-unglaubbut it's still no problem for the file checksums that a normal package has09:02
diverseso one of the ideas Romster had was implementing usage of sha512 for file checksums09:03
*** phant0mas has quit IRC09:04
diverseso I guess what I could do is support a .sha512sum file for checking?09:04
leo-unglaubyes09:04
Romsterwell i was using sha256 at one stage09:04
Romsterbecause sha1 is broken09:04
leo-unglaubi would vote for sha51209:05
leo-unglaubif crux wants to become the reference distro for security09:06
leo-unglaubgo for sha51209:06
diversewell try to influence the developers, in the meantime I still have to support md5 for backwards compatibility09:06
Romsterhttps://crux.nu/bugs/index.php?do=details&task_id=223&string=sha256&project=1&search_name=&type%5B0%5D=&sev%5B0%5D=&pri%5B0%5D=&due%5B0%5D=&reported%5B0%5D=&cat%5B0%5D=&status%5B0%5D=open&percent%5B0%5D=&opened=&dev=&closed=&duedatefrom=&duedateto=&changedfrom=&changedto=&openedfrom=&openedto=&closedfrom=&closedto=09:07
Romsterbut it wont happen our use case says its fine for what we use it for.09:07
leo-unglaubi could disagree with that *g*09:08
diverseleo-unglaub: but what I could do is priority check for sha512 first and then check for md5 ;)09:09
Romstergo add to it then.09:09
leo-unglaubRomster: no, sepen and i disagree most of the time *g*09:09
Romsterdiverse, that isn't a bad idea.09:09
leo-unglaubdiverse: sure, check both09:09
leo-unglaubprefer sha51209:10
leo-unglaubif not existant, throw a warning an use md409:10
leo-unglaubäh, md509:10
diversehehe09:10
Romstermd4 is still used actually.09:10
Romsterccache uses md409:10
diverseleo-unglaub: well, sepen is not around all that often anymore, so... :P09:10
leo-unglaubOMFG .... they use md4???09:11
Romsteryeah it needs to be extremely fast09:11
leo-unglaubstill ... from a compiler cache i expect to be 100% collition free ..09:12
leo-unglaubtraching down that bug ...09:12
RomsterMD4 is nowadays too weak to be useful in cryptographic contexts, but it should be safe enough to be used to identify recompilations.09:12
Romsterman ccache09:12
b1n0:( cant solve a problem with directfb, it doesnt detect keyboard (input devices)09:12
Romsterleo disagres with i2p too *runs for cover*09:13
diversehehehehe09:13
Romsteryeah md4 is fast i guess for ccache use09:13
diversebut the point leo was making that sha2 has improved software wise that performance shouldn't be a problem09:14
Romsterah09:15
diversethus why I asked him earlier about sha109:15
diverseand since md5 has been around a long time as well, at this point the margin difference between md4 and md5 are probably microscopically minor09:18
diverseleo-unglaub: did I assume correctly?09:19
leo-unglaubwell, they added a new round and some functions have changed the accepted params order09:24
leo-unglaubthat way they look more random09:25
diversein relation to sha2?09:26
leo-unglaubno, just md4 -> md509:26
diverseoh I see, so not much operation between the two09:26
leo-unglaubno09:27
Romsterif you can hash faster than md4 and be secure i'd be interested09:28
Romsterheck even if it uses cuda09:28
leo-unglaubhaha, cuda would be cheating *g*09:29
diversewell the point is, you aren't benefitting all that much with md4 compared to md5 anyway09:29
*** kyubiko_school has joined #crux09:30
leo-unglaubspecially in ccache09:30
leo-unglaubif you have large codebases like firefox, ...09:30
leo-unglaubjust one collision and the build fails09:30
leo-unglauband finding that bug is hard09:30
diverseah ha :)09:31
diverseso that's where are the trouble starts09:31
diversethey should really start accepting md5, if not at least via autotool flag09:32
*** doomicide has joined #crux09:32
diversea/are/all/09:33
leo-unglaubthey should .. but good luck getting stuff fixed on that low level09:33
diversethat doesn't give me a whole lot of hope :(09:34
diverseleo-unglaub: do you think you will be able to give me a hand in metaport?09:36
diverseoh right... I forgot...09:37
leo-unglaubproject ...09:37
leo-unglaub:(09:37
diversegithub isn't your cup of tea09:37
Romsterno i thnk leo is unable to commit... see what did there09:40
leo-unglaub?09:40
Romsterproject. commitment... git commit09:41
diverseoh I thought "project" meant the thing he is currently working on for his job?09:41
leo-unglaubyes, it meant the project in my company09:41
leo-unglaubthat keeps me from doing other stuff09:41
Romsteroh i thought you mean diverse's project.09:42
leo-unglaubnooooooo09:42
Romsterguess i interpreted that wrongly.09:42
diversehmm, I see a cleanup optimization opportunity09:47
diversebrb, need food10:04
*** hhhhhhhh has joined #crux10:13
*** b1n0 has quit IRC11:06
*** GuY` has joined #crux11:14
*** fnords has quit IRC11:21
*** fnords has joined #crux11:21
*** xeirrr has joined #crux12:32
*** jdolan has quit IRC12:37
*** hhhhhhhh has quit IRC12:38
*** hhhhhhhh has joined #crux12:38
*** jdolan has joined #crux12:38
*** jdolan has quit IRC12:51
*** xeirrr has left #crux ()13:18
*** dxlr8r has joined #crux13:23
*** phant0mas has joined #crux13:26
*** hhhhhhhh has quit IRC13:31
*** phant0mas has quit IRC13:47
*** pejman has quit IRC14:06
*** doomicide has quit IRC14:09
*** hhhhhhhh has joined #crux14:09
*** pejman has joined #crux14:13
*** pejman has quit IRC14:13
*** pejman has joined #crux14:13
*** hhhhhhhh_ has joined #crux14:28
*** _root_ has joined #crux14:36
*** jdolan has joined #crux14:49
diversekori: in your driver you can change that fetch to a pull again15:01
diversealso change the "-q" to "-v" if you haven't15:10
diverse(debugging is important at this stage)15:11
*** leo-unglaub has quit IRC15:14
*** dougiel has joined #crux15:38
*** phant0mas has joined #crux15:38
*** BitPuffin has joined #crux15:49
*** hhhhhhhh_ has quit IRC16:02
*** hhhhhhhh has quit IRC16:07
*** EmoSpice has joined #crux16:25
*** leo-unglaub has joined #crux16:27
*** arduo has joined #crux16:36
*** doomicide has joined #crux16:42
*** lnds has joined #crux16:59
*** kori has quit IRC17:21
*** kori has joined #crux17:24
*** kori has quit IRC17:30
*** leo-unglaub has quit IRC17:43
*** hhhhhhhh has joined #crux17:47
*** leo-unglaub has joined #crux18:06
*** kori has joined #crux18:18
*** kori has quit IRC18:37
*** kori has joined #crux18:38
*** kori has quit IRC18:38
*** kori has joined #crux18:38
*** kori has quit IRC18:51
*** SiFuh has joined #crux18:57
*** kori has joined #crux19:02
*** _root_ has quit IRC19:32
koridiverse: aye, I think I'm going to keep git fetch19:37
korigit is very flexible19:37
koriso we can make the git driver a tiny bit more complex, and be able to set how you want to proceed every time when syncing with ports -u19:38
koriI'm not sure if that's ideal19:38
koribut it's possible19:38
*** leo-unglaub has quit IRC19:40
*** heroux has joined #crux19:41
rmullRomster: Do you happen to know off the top of your head if the hash function in ccache can be swapped out for another without uprooting too much stuff? Might be fun to experiment with alternatives. I see some non-cryptographic fast hashes posted online. I'm sure someone's already looked into it though19:47
*** kyubiko_school has quit IRC20:00
*** kori has quit IRC21:01
*** kori has joined #crux21:03
*** phant0mas has quit IRC21:13
*** Feksclaus has joined #crux21:14
*** doomicide has quit IRC21:19
*** lnds has quit IRC21:24
*** gawd has joined #crux21:59
*** jdolan has quit IRC22:01
*** Pingax has quit IRC22:08
Worksterrmull, no idea i never investigated22:32
Worksteryou should add a feature request to ccache and show them these new fast hashes22:33
*** jdolan has joined #crux22:39
*** gawd has quit IRC22:42
prologicWhooo23:18
prologicDocker 1.5 is coming soon23:18
prologicRC is out23:18
prologiclots of new features :)23:18
prologicIPv6 :) Swarm, Machine and Compose23:18
prologicwoot :)23:18
*** dougiel has quit IRC23:32

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!