| 00:00.51 | *** join/#oe chouimat|ibook (~dieu@r2351064.cidc.net) |
| 00:01.40 | *** join/#oe chouimat (~dieu@r2351064.cidc.net) |
| 00:21.05 | pb_ | kergoth: in the current implementation? shlibdeps is in package.bbclass, not debian. |
| 00:21.14 | pb_ | all that debian.bbclass does is set the policy for library naming. |
| 00:21.41 | kergoth | ahh right |
| 00:27.16 | CosmicPenguin | I wonder why AC_CHECK_FILE is illegal for xcompiling |
| 00:27.21 | treke | will shlibdeps work with redhat naming though? |
| 00:27.33 | treke | I thought something broke in redhat naming |
| 00:28.03 | pb_ | yeah, it should be fine |
| 00:28.33 | pb_ | I think there are a few packages that have hard-coded debian style names in their RDEPENDS, but not many. |
| 00:29.04 | kergoth | should really have a convenience function to call out to do the name mangling explicitly for those packages |
| 00:32.00 | pb_ | the trouble is that the name mangling is contingent on the (contents of the) files in the package. |
| 00:32.25 | pb_ | you can't easily say "tell me what package <blah> would rename to", without having a copy of that package on hand to look at. |
| 00:33.34 | kergoth | ah, good point |
| 00:34.05 | *** join/#oe CoreDump|hom (~mhentges@xdsl-213-196-245-40.netcologne.de) |
| 00:34.14 | pb_ | the only thing we could do would be to record a mapping table somewhere in staging. that way, as long as the depended-on package had been built before the one doing the depending, you could figure it out. |
| 00:34.31 | treke | that sounds like a safe assumption |
| 00:35.25 | pb_ | as long as you adhere to the "if in RDEPENDS, also in DEPENDS" credo, yeah |
| 00:35.56 | pb_ | I'm not sure if that's still controversial. |
| 00:36.19 | pb_ | well, I guess we could require it in this specific case anyway |
| 00:36.32 | treke | are there cases where shlibdeps could find a dependency that wouldnt actually be required to build a package? |
| 00:36.53 | pb_ | no; none of this applies to shlibdeps. |
| 00:37.01 | treke | ah. then ignore me |
| 00:37.07 | pb_ | shlibdeps, by definition, will only find dependencies on packages that have already been built, and will always get the names right. |
| 00:37.44 | pb_ | the case that causes trouble is where a .bb file explicitly mentions some random package in its RDEPENDS. |
| 00:38.05 | kergoth | the most common case is that of an interpreter. |
| 00:38.10 | kergoth | which isnt necessarily required to build the package |
| 00:38.13 | kergoth | only to run it |
| 00:38.17 | pb_ | yeah, exactly |
| 00:39.11 | pb_ | my contention has always been that, since you need some way to ensure that the interpreter has been built by the time you come to do image construction, you should put it in DEPENDS even though it isn't involved in the act of building your package. |
| 00:39.39 | treke | fair point |
| 00:39.52 | pb_ | otherwise you'll wind up in do_rootfs with this dangling RDEPENDS, and oe will have no clue how to make a package to fulfil it. |
| 00:42.00 | Mech422 | kergoth: that article any good ? |
| 00:42.05 | Mech422 | lamson, that is |
| 00:42.46 | *** join/#oe chouimat (~dieu@r2351064.cidc.net) |
| 00:44.11 | treke | where did oe commander go? |
| 00:44.35 | Mech422 | ~ oecommander |
| 00:44.41 | Mech422 | blah |
| 00:44.47 | Mech422 | treke: whats an oe commander ? |
| 00:45.03 | bluelightning | one above an oe captain? :) |
| 00:45.23 | treke | ah here we go, we still have the oe.deprecated repo I can pull it from |
| 00:45.39 | Mech422 | bluelightning: :-) |
| 00:45.39 | treke | Mech422: an oe front end. |
| 00:46.07 | Mech422 | treke: is there someplace I can RTFM about it ? |
| 00:46.14 | treke | no |
| 00:46.17 | pb_ | treke: I think mickey pulled it from the repo to stop it going into bitbake 1.0 |
| 00:46.26 | *** join/#oe CIA-4 (~CIA@to.je.spocco.com) |
| 00:46.40 | treke | yeah just wanted to make sure the source didn't completely disappear |
| 00:50.11 | bluelightning | who is in charge of hotplug in oe? |
| 00:50.37 | bluelightning | (no maintainer in the .bb) |
| 00:51.35 | pb_ | I think kergoth looks after it, mostly. |
| 00:53.19 | CosmicPenguin | lack of maintainer implies kergoth |
| 00:53.22 | CosmicPenguin | or pb_ |
| 00:55.19 | bluelightning | ok |