"bzr log" by revision-id is slow
Eli Zaretskii
eliz at gnu.org
Wed Oct 12 22:18:24 UTC 2011
It looks like "bzr log -r REVID" is O(1), but with a very large
constant. The time is roughly the same as it takes "bzr
revision-info" to return the revno by revision-id.
Here are a few examples from the Emacs repository, which show that the
time taken by "bzr log -r REVNO" grows for older revisions, but the
same command by REVID takes constant time, which happens to be the
same as for "bzr log -r 1".
Is this expected?
D:\gnu\bzr\emacs\trunk>timep bzr log -r-2
106069: Stefan Monnier 2011-10-12 * doc.c (get_doc_string): Encode file name.
real 00h00m00.625s
user 00h00m00.390s
sys 00h00m00.187s
D:\gnu\bzr\emacs\trunk>bzr lookup-revision 106069
monnier at iro.umontreal.ca-20111012190804-frm225b6jckvaj7k
D:\gnu\bzr\emacs\trunk>timep bzr log -r monnier at iro.umontreal.ca-20111012190804-frm225b6jckvaj7k
106069: Stefan Monnier 2011-10-12 * doc.c (get_doc_string): Encode file name.
real 00h00m19.140s
user 00h00m18.640s
sys 00h00m00.421s
D:\gnu\bzr\emacs\trunk>timep bzr log -r cyd at stupidchicken.com-20090609183929-nuq8sa1d8pep960t
96032: Chong Yidong 2009-06-09 * ada-mode.texi (Installation, Compile commands)
real 00h00m19.265s
user 00h00m18.796s
sys 00h00m00.390s
D:\gnu\bzr\emacs\trunk>timep bzr revision-info cyd at stupidchicken.com-20090609183929-nuq8sa1d8pep960t
96032 cyd at stupidchicken.com-20090609183929-nuq8sa1d8pep960t
real 00h00m19.187s
user 00h00m18.828s
sys 00h00m00.296s
D:\gnu\bzr\emacs\trunk>timep bzr log -r 96032
96032: Chong Yidong 2009-06-09 * ada-mode.texi (Installation, Compile commands)
real 00h00m02.750s
user 00h00m02.406s
sys 00h00m00.312s
D:\gnu\bzr\emacs\trunk>timep bzr log -r 1
1: Jim Blandy 1985-04-18 entered into RCS
real 00h00m20.828s
user 00h00m19.218s
sys 00h00m00.609s
D:\gnu\bzr\emacs\trunk>timep bzr log -r jimb at redhat.com-19850418004829-3xjfpek0kofyaf5q
1: Jim Blandy 1985-04-18 entered into RCS
real 00h00m19.968s
user 00h00m19.343s
sys 00h00m00.546s
More information about the bazaar
mailing list