Merge lp:~jameinel/bzr-fastimport/mooloolaba into lp:~bzr/bzr-fastimport/fastimport.dev

Proposed by John A Meinel
Status: Rejected
Rejected by: Ian Clatworthy
Proposed branch: lp:~jameinel/bzr-fastimport/mooloolaba
Merge into: lp:~bzr/bzr-fastimport/fastimport.dev
To merge this branch: bzr merge lp:~jameinel/bzr-fastimport/mooloolaba
Reviewer Review Type Date Requested Status
Ian Clatworthy Disapprove
Review via email: mp+15365@code.launchpad.net

This proposal supersedes a proposal from 2009-11-15.

To post a comment you must log in.
Revision history for this message
John A Meinel (jameinel) wrote : Posted in a previous version of this proposal

This branch is in --2a format, but trunk isn't, so this request may fail.

Anyway, this is just a rollup of the stuff I worked on in Mooloolaba. Stuff like:

1) Memory reduction by referencing texts as strings rather than lines. This alone caused my peak qt import memory to go to 1.5GB from >2GB (I don't know how big, because I go OOM @2GB on my 32-bit machine).
2) Memory reduction by not caching all texts in memory. Instead, dump them into temp files. 'small' blob files are all collected into a single temp file, everything else gets a unique file that will be deleted when the last reference goes away. This got me down to between 500MB => 900MB during the import.
3) Switch the code that adds things to the repo to use CommitBuilder. It cheats a little bit here and there to get around deficiencies in the api. Namely, CommitBuilder wants to start_write_group and commit_write_group for every build, and we wanted to work around that.

I'm not sure what else is included, but that is the basics.

Revision history for this message
Ian Clatworthy (ian-clatworthy) wrote : Posted in a previous version of this proposal

Can you push this branch again now that the trunk is in 2a format please?

Revision history for this message
John A Meinel (jameinel) wrote :
Revision history for this message
Ian Clatworthy (ian-clatworthy) wrote :

https://code.edge.launchpad.net/~jameinel/bzr-fastimport/mooloolaba reports "This branch has not been pushed yet."

Revision history for this message
Ian Clatworthy (ian-clatworthy) wrote :

OK, looks like a proper merge proposal supersedes this one.

review: Disapprove
Revision history for this message
John A Meinel (jameinel) wrote :

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Ian Clatworthy wrote:
> https://code.edge.launchpad.net/~jameinel/bzr-fastimport/mooloolaba reports "This branch has not been pushed yet."

It was pushed, but it looks like the stacking format difference bugs
caused problems. I think it currently thinks it is a lightweight
checkout of C:/... which is a bit wonky. Anyway, as you noticed, I
pushed again to a different branch.

John
=:->

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAksZIqoACgkQJdeBCYSNAAPz3QCgwz19DAdQh79OQM0Cfj2gWVbG
qecAnjTffVo4JtDkL74l6pVK/kYJGSRD
=2ikp
-----END PGP SIGNATURE-----

Subscribers

People subscribed via source and target branches

to all changes: