IRC logs for #baserock for Saturday, 2014-10-25

*** zoli__ [~zoli_@0x5e91887a.adsl.cybercity.dk] has quit [Remote host closed the connection]00:12
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock08:00
*** dabukalam [~quassel@87.113.195.160] has joined #baserock09:31
*** dabukalam [~quassel@87.113.195.160] has quit [Remote host closed the connection]09:36
*** dabukalam [~quassel@87.113.195.160] has joined #baserock09:37
*** dabukalam [~quassel@87.113.195.160] has quit [Client Quit]09:40
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]09:40
*** dabukalam [~quassel@87.113.195.160] has joined #baserock09:40
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock09:48
*** dabukalam [~quassel@87.113.195.160] has quit [Remote host closed the connection]10:04
*** dabukalam [~quassel@87.113.195.160] has joined #baserock10:04
*** dabukalam [~quassel@87.113.195.160] has quit [Ping timeout: 260 seconds]10:10
*** dabukalam [~quassel@87.113.195.160] has joined #baserock10:13
*** dabukalam [~quassel@87.113.195.160] has quit [Client Quit]10:16
*** dabukalam [~quassel@87.113.195.160] has joined #baserock10:16
*** dabukalam [~quassel@87.113.195.160] has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]10:22
*** dabukalam [~quassel@87.113.195.160] has joined #baserock10:23
jjardon_rjek: the patch would be to upgrade the version of readline, the version in baserock is ancient (for license reasons) and it cause compilation errors when trying to compile a modern version of nm10:30
*** paulsherwood87 [~paulsherw@cpc65054-bagu12-2-0-cust948.1-3.cable.virginm.net] has joined #baserock10:32
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]10:40
*** paulsherwood87 [~paulsherw@cpc65054-bagu12-2-0-cust948.1-3.cable.virginm.net] has quit [Ping timeout: 265 seconds]10:42
straycatFor the moment you could make a genivi version of the stratum that contains readline, keeping it at its current version. Make the genivi systems depend on that stratum. Then upgrade the version of readline everything else uses.10:51
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock10:51
*** dabukalam [~quassel@87.113.195.160] has quit [Remote host closed the connection]11:19
*** dabukalam [~quassel@87.113.195.160] has joined #baserock11:21
jjardon_straycat: that would not work, you would need to duplicate all the strata on top of core11:22
*** dabukalam [~quassel@87.113.195.160] has quit [Client Quit]11:22
*** dabukalam [~quassel@87.113.195.160] has joined #baserock11:23
jjardon_mmm, maybe we can create a readline-common and readline-common-genivi stratum, make core depend on readline-common and then in the genivi systems use readline-common-genivi instead readline-common11:25
*** dabukalam [~quassel@87.113.195.160] has quit [Client Quit]11:25
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has joined #baserock11:28
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has quit [Client Quit]11:30
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has joined #baserock11:31
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has quit [Client Quit]11:31
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has joined #baserock11:32
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has quit [Client Quit]11:33
*** dabukalam_ [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has joined #baserock11:34
*** dabukalam_ [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has quit [Client Quit]11:34
straycatjjardon_, *nod* that's implied by making the genivi system depend on a genivi core11:34
*** zoli_ [~zoli_@linaro/zoli] has quit [Read error: Connection reset by peer]11:38
*** zoli__ [~zoli_@0x5e91887a.adsl.cybercity.dk] has joined #baserock11:38
rjekjjardon_: I mean patch nmcli so it doesn't need the latest readline.11:44
rjekOr patch it so it doesn't need it at all!11:44
rjekGENIVI is not the only project that is allergic to GPLv3.11:44
straycatIf anyone actually wants to use it they might want readline11:44
rjekNobody *wants* to use nmcli :)11:45
*** zoli__ [~zoli_@0x5e91887a.adsl.cybercity.dk] has quit [Ping timeout: 256 seconds]11:48
jjardon_rjek: no, the correct fix is to find a solution so everyone can use a modern readline in baserock, no patch every module that can have problems with a very old and probably unmaintained readline11:51
jjardon_straycat: thinking twice I do not think what I said before would work: every program compiled against the modern readline will complain in runtime if the old readline is present instead11:54
straycatMaking genivi systems depend on a genivi core that includes the current readline would work. As you say you'd also have to duplicate core's dependants, it's not necessarily a nice solution but it should work.11:59
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has joined #baserock12:03
jjardon_straycat: are you suggesting to have core a genivi-core? that means we have to duplicate all the strata on top! foundation-genivi, mesa-common-genivi, llvm-common-genivi, grpahics-common-genivi .... and all of them with exactly the same content; it would work but I do not think is a valid solution12:04
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has quit [Client Quit]12:06
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has joined #baserock12:07
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has quit [Client Quit]12:08
*** dabukalam [~quassel@ec2-54-69-244-150.us-west-2.compute.amazonaws.com] has joined #baserock12:08
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock12:12
straycatWell if it works, it's valid imo :p But yes I agree that we don't really want to be duplicating all that stuff.12:12
straycatIt seems to me that morph ought to be able to handle this case by providing a way to conditionally select chunks.12:14
straycatThat said, we are deliberately trying to keep definitions simple.12:15
*** zoli_ [~zoli_@linaro/zoli] has quit [Ping timeout: 240 seconds]12:37
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock12:43
rjekjjardon_: Then the solution is to relicence readline under GPLv3 :)12:44
rjekreadline has basically been static for decades.  I wonder what they changed that nmcli depends on12:45
rjekStill, who's a version junky in this day and age?12:45
*** zoli_ [~zoli_@linaro/zoli] has quit [Ping timeout: 272 seconds]12:51
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock12:53
radiofreejjardon_: network manager isn't in a genivi system right?12:56
radiofreecould you not just add readline to the network manager stratum?12:56
straycatWe build python with readline so it needs to be in core at the moment.12:59
radiofreei was thinking you add it as a chunk in NM, make it build depend on that13:01
radiofreebut then i suppose there might be problems when you install it13:02
radiofreesince the old readline that python etc.. were built against will be replaced with the newer one13:02
jjardon_radiofree: exactly13:02
radiofreecould you just not build readline?13:03
radiofreesorry, nmcli13:03
jjardon_radiofree: no, readline is a hard dependency of NM13:06
radiofreemaybe you should submit a patch for that then13:07
jjardon_no, as I said before, the correct fix is to find a solution so everyone can use a modern readline in baserock, no patch every module that can have problems with a very old and probably unmaintained readline13:08
radiofreeso what solution are you proposing/considering for that?13:09
jjardon_now is readline, in the future it can be any other GPLv3 module13:09
jjardon_as straycat said, I guess morph should  handle this case by providing a way to conditionally select chunks. This will fix the dependency problems when you want to build wayland-only systems as well13:12
radiofreewell customers who might want to use baserock (or any type of embedded system) are gplv3 averse, so imo it's a mistake on network managers part to do that13:12
radiofreeunless they're totally resigned to never being used in an embedded context anyway because of connman13:13
jjardon_radiofree: Maybe, but NM problem is a specific use case of a general problem baserock has 13:15
jjardon_jhbuild solves this with the "--conditions" argument: https://developer.gnome.org/jhbuild/stable/command-reference.html.en , example of use: https://git.gnome.org/browse/jhbuild/tree/modulesets/gnome-suites-core-deps-3.16.modules#n10713:47
straycatI expect we'd want the conditions to be defined in the definitions rather than on the command line.13:49
jjardon_straycat: they are defined in the "definitions" (check the moduleset file I pointed in the second link)13:52
* straycat nods13:53
straycatI was commenting on the --conditions flag13:53
straycatI like their dot command more than our graph commands14:06
jjardon_straycat: yeah, dot is very cool! is it possible to do the same in baserock?14:20
jjardon_didnt know that14:20
straycatjjardon_, morph has a graph-build-depends command at the moment, but it's not useable because it graphs an entire system and even for our smallest system the entire build graph is too large to be graphed. I've been working on two new commands to replace graph-build-depends, those changes aren't finished yet and I might tweak them further now that I've seen jhbuild's dot.14:29
rjekIt is a bug for something to depend on readline.14:33
rjekSubmit a patch upstream to remove the dependancy.14:33
jjardon_rjek: I think you are missing the point here: the bug is in baserock, where I can not use a specific version of a component because morph doesnt have a way to conditionally select chunks14:45
rjekOh yes, we want that.14:45
rjekBut also, I don't want readline in any embedded system I build.14:46
jjardon_rjek: why do you think is a bug to depend on readline?14:46
rjek(which is why http://bzr.rjek.com/public/fakeline exists)14:46
*** jjardon [~jjardon@82-70-136-246.dsl.in-addr.zen.co.uk] has quit [Remote host closed the connection]14:46
jjardon_ is now known as jjardon14:46
rjekjjardon: Because it's totally unnecessary?14:46
rjekWhen building something that has to fit in a few tens of MB (or even a few MB), the ~300kB of bloat and likely as-yet-undiscovered environment security bugs isn't welcome for something in the typical case nobody will be typing anything into.14:48
rjekAlso, it's possible somebody would want to use Network Manager in a non-GPLv3 system.14:50
rjekI bet it would be trivial to remove the dependancy on the newer version of readline.  It's probably even accidental.14:50
jjardonrjek: currently readline is in baserock because python, no nm14:51
rjekThere were historically other things that hard-depended on it.14:51
rjekBut you're missing my point.14:51
jjardonrjek: and btw, that kind of system (fit in tens of MB) doesnt exist in current baserock definitions14:56
rjekbaserock isn't the examples it comes with :)14:56
jjardonrjek: you can say the stuff in system/ are examples, I wouldnt say the same about strata/15:00
paulsherwoodso afaik readline is only used for commandline processing15:04
paulsherwoodgenivi systems should not have any commandline processing going on15:04
paulsherwoodbut i think we've been around this topic quite a few times15:05
paulsherwoodjjardon: all of the strata are examples too i think15:06
jjardonpaulsherwood: not sure I agree: the metada that defines how to build all the stuff is not only examples; then we can say yocto is bitbake + some examples as well15:17
straycatDoes that mean our releases are examples?15:24
paulsherwoodjjardon: yes, that's what i mean. as a user, it's much easier to create something if i have existing examples to work from15:25
paulsherwoodstraycat: i don't understand your question15:25
paulsherwoodhowever i have one of my own...15:26
paulsherwoodthis new 'build-system' - what's it for?15:27
straycatI don't see how we can say that the systems we use every day for development are examples.15:27
straycatI'm not saying they can't be used as examples though15:28
paulsherwoodstraycat: that's my point. saying they are not examples seems to deny that possibility15:29
straycatThis is semantic anyway because they are examples but they're also more than examples because they're the systems we use daily.15:29
paulsherwoodi'm not saying they're *only* examples. but they are examples.15:30
* straycat nods15:30
* paulsherwood attempts to re-phrase on build-system15:31
paulsherwoodshould it be possible to run a build-system for normal morph builds?15:31
paulsherwoodie could i use build-system instead of devel?15:31
straycatI think so, it's just a devel system with extra stuff required for distbuild.15:33
* paulsherwood notices that it seems to insist on various config items related to distbuild...15:38
* straycat nods15:39
straycatyou'd still need to deploy it as a distbuild node, but there's no reason you can't ssh in after that and run local builds15:39
jjardonis it ok to ask here for committing patches like this?  http://fpaste.org/145139/41425147/ (Upgrade a version in the same stable branch so no API breaks)15:42
paulsherwood+1 from me15:42
paulsherwood(assuming you've built it)15:42
jjardonthanks!15:44
jjardoncan I commit directly to master or should I create a branch and a  merge commit?15:45
paulsherwoodi think technically we're still operating on +2, so you may need a comment from straycat or someone else15:47
paulsherwoodi don't know what the policy is on merge commits15:49
straycatSeems sane to me, so if you've tested it then it should be fine.15:49
jjardongreat, thanks!16:04
jjardonCould someone also take a look to the "Clean some genivi-related strata" series I sent a while ago? Cheers!16:04
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]16:17
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock16:17
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]17:06
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock17:06
straycatcan't just now sorry :/17:15
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]18:09
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock18:10
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]18:25
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock18:25
* straycat discovers that drafting emails can be a surpsingly good way to force yourself to think about both sides of a technical discussion19:43
* straycat deletes draft19:43
paulsherwoodi hope that doesn't mean you've just decided to ignore reasonableness and stick to your guns :)19:52
straycathaha!19:53
* paulsherwood builds baserock/jjardon/clean_genivi on x86_64 and jetson19:55
paulsherwoodstraycat: do you know if it's straightforward to run a baserock arm system on x86 using qemu?20:14
* paulsherwood notices http://wiki.baserock.org/genivi/#index4h220:18
rjekEvening.21:29
paulsherwoodhi21:32
* pedroalvarez waves21:34
* straycat meows21:34
radiofreepaulsherwood: it's possible but will be very slow21:41
pedroalvarezIndeed, and you have to use the versatile bsp to follow that instructions 21:46
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]22:08

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!