[07:50] *** WeaselSoup has left "WeeChat 2.1"
[11:34] *** iridium has quit IRC (Ping timeout: 240 seconds)
[14:13] *** ali1234 has left 
[17:47] <ali1234> i need to build a driver with media_build
[17:47] <ali1234> i don't want to build all of them though
[17:47] <ali1234> it takes too long
[17:47] <ali1234> so how do i limit what it builds?
[17:47] <ali1234> i gather it has something to do with 00_opt_default.inc
[20:48] <nst> mchehab: ping, gentle reminder/annoyance on the https://patchwork.linuxtv.org/patch/49403/ series to finish up ddbridge-0.9.33, guess https://patchwork.linuxtv.org/patch/49456/ won't work out though
[21:18] <mchehab> nst, what do you mean by "won't work out"?
[21:18] <nst> mchehab: "won't work out" as in "too late", "needs thorough review" or "needs changes"
[21:20] <nst> though ralph (rjkm) and me already agreed on the different ioctl ids and he'll pick them up into upstream
[21:20] <mchehab> ok, I'll postpone they then
[21:22] <nst> well, if the patches are ok for you, it'd be great to get them into 4.18 as they could be really useful (required) regarding fpga updating, especially as dd pushed out updates for the cines2v7 line, and the mci cards somewhat rely on them
[21:23] <nst> (and its another topic to whipe off the todo list)
[22:43] <b-rad> ali1234, in theory you might be able to use the new config-mycompat.h to blacklist entire driver families to prevent building them
[22:44] <b-rad> aka undfine CONFIG_RC_CORE to not build that entire class, etc
[22:45] <b-rad> when i made it i didn't think of using it like that, but I think it would work the same
[22:45] <ali1234> hmm
[22:45] <ali1234> the main problem is that dkms takes 2 seconds to delete an installed module, and media_build installs 600+ modules
[22:45] <b-rad> if you want to try then make config-mycompat.h in v4l/ and put your #undefine's there
[22:46] <b-rad> thats a feature, not a problem
[22:47] <b-rad> i recently added facility to override that and just told you how, so try
[22:47] <ali1234> i dont know if i really want to remove all the things
[22:47] <ali1234> or fix dkms
[22:48] <ali1234> make: *** No rule to make target 'config-mycompat.h'. Stop.
[22:48] <ali1234> or you mean make as in make it with a text editor?
[22:50] <ali1234> the string "mycompat" doesn't exist anywhere in either my kernel tree or my media_build tree
[22:50] <b-rad> you create that file externally, the copy it into v4l/ and it will be suckedi into the build process and override anything desired
[22:50] <b-rad> read v4l/compat.h
[22:51] <b-rad> my description was as explicit as i thought dictated, if you think i can make it clearer please suggest changes
[22:52] <b-rad> the description might not apply to you directly, but if you #undefine kernel config options i do think it will operate the same way due to how it is included
[22:53] <b-rad> and it not, excuse my distraction, and do carry on lol
[22:53] <b-rad> *if not*
[22:54] <ali1234> yeah my compat.h does not include config-mycompat.h
[22:54] <b-rad> well then it's out of date
[22:57] <ali1234> all the patch does is include that file?
[22:58] <b-rad> pretty much, yes
[22:58] <b-rad> makes & cleans it too, but yes
[22:59] <b-rad> just be sure you do the include at the same spot, so your overrides go into effect
[22:59] <ali1234> i will cherry pick this commit
[23:00] <ali1234> but a question: you wrote: "should be saved externally and copied into v4l/ when required."
[23:00] <b-rad> do let me know how it works out, you'll be the second user lol
[23:00] <ali1234> when is that?
[23:00] <ali1234> based on https://github.com/ali1234/tbs-dkms/blob/master/doit.sh
[23:00] <b-rad> save it anywhere in the system you like, just not in v4l/, because distclean will delete it
[23:00] <ali1234> when is distclean run?
[23:01] <b-rad> just like if it does not exist during the initial stages an empty version will be created
[23:01] <b-rad> when you run it
[23:01] <ali1234> i have but i don't run it
[23:01] <b-rad> look do what you want but you'll be the one screaming when you do accidentally distclean and hadn't saved that file elsewhere
[23:01] <ali1234> it will be checked in to git
[23:01] <ali1234> it will just check it out again
[23:02] <b-rad> just realize if you distclean it will disappear
[23:02] <b-rad> and do let me know if it works like you want, i'm happy you're using it
[23:02] <ali1234> my question is this: do any other targets run distclean?
[23:03] <ali1234> or can it only be run explicitly?
[23:03] <b-rad> i cannot answer, because when i use that system i do the stages myself explicitly
[23:03] <b-rad> i have my own super system per se
[23:03] <ali1234> so do i
[23:03] <ali1234> i need to know if it is going to run "make distclean" at an unexpected moment
[23:03] <b-rad> i don't think so
[23:04] <b-rad> i put it there to be as least destructive as possible
[23:04] <ali1234> if i want to clean i use "git clean -d -f -x"
[23:04] <b-rad> well i change architectures, patch levels, driver options, etc
[23:04] <b-rad> so i use distclean
[23:06] <b-rad> just try it and see
[23:07] <ali1234> i now have a script that builds a debian package script that builds a dkms deb that uses media_build that uses the kernel makefiles to build linux_media... that's five levels of meta
[23:09] <ali1234> every time i build a new version of the package it takes half an hour to uninstall the old one :(
[23:11] <b-rad> turtles all the way down
[23:34] <ali1234> okay, it didn't work
[23:34] <ali1234> i did #undef CONFIG_VIDEO_ZORAN but it is still set to =m in the resulting .config
[23:39] <ali1234> i dont think there are really that many modules i can safely remove... so i guess i have to fix dkms