[07:31] <Pali> hello, I have problem with importing svn repo into launchpad bzr (via launchpad auto import service)... for 5 days it trying to import svn repo every 30 minutes and something is not working
[07:31] <Pali> bzr repo is still empty
[07:31] <Pali> it is normal?
[07:33] <Pali> here is repo: https://code.launchpad.net/~pali/llvm/llvm-3.5
[07:33] <wgrant> Pali: Subversion imports are done incrementally, a few thousand revisions at a time.
[07:33] <wgrant> Something like LLVM has hundreds of thousands of revisions of history; it will take many attempts before it completely imports.
[07:33] <wgrant> Ah, 500 revisions at a time, in fact.
[07:35] <Pali> wgrant: so I just need wait more time?
[07:35] <wgrant> Pali: Right.
[07:35] <wgrant> Importing a large Subversion history can be very slow due to the nature of the protocol.
[07:35] <Pali> ok
[07:37] <Pali> looks like llvm has 216045 revisions, so it will take about 216045/500/2/24 ~= 9 days?
[07:39] <Pali> wgrant: bzr repos from svn branches do not support stacked bzr?
[07:39] <Pali> because there is already autoimporting llvm trunk branch into bzr (from svn), but not branch 3.5
[07:40] <Pali> and both branches have big same history
[07:40] <wgrant> Pali: They do not.
[07:40] <wgrant> Imports are always independent.
[07:43] <Pali> ok
[07:44] <Pali> thanks for info, I will wait 4 days and if there will be problem I ask again
[07:44] <Pali> wgrant: another question: it is possible to check how many revs are imported now?
[07:48] <wgrant> Pali: No, you'd have to estimate from the number of imports so far.
[07:48] <Pali> ok
[11:46] <jelmer> wgrant: imports should be stacked I think
[11:46] <jelmer> wgrant: or did I have a branch for that that never landed?
[11:54] <wgrant> jelmer: I know you had an incomplete branch at some point. I don't remember how far it got, except that it certainly doesn't work on the side of the importer.
[11:55] <wgrant> They may be stacked on codehosting, but they're certainly independent on the importd intermediary.
[11:55] <wgrant> And the latter is all that matters for performance.
[11:55] <jelmer> wgrant: ah, yeah. That must be it.
[11:55] <jelmer> codehosting does stacking, and my branch to use stacking on the importer never landed.
[11:56] <jelmer> (I remember we had bugs with regards to stacking for code imports)