Bug 280211 – Add support for EBZR_OFFLINE and ESCM_OFFLINE to bzr.eclass

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Jonathan Callen (RETIRED) 2009-08-03 20:39:17 UTC]

This patch adds support for EBZR_OFFLINE and ESCM_OFFLINE to bzr.eclass, in a method similar to git.eclass and svn.eclass.

[Comment 1 Jonathan Callen (RETIRED) 2009-08-03 20:44:03 UTC]

Created attachment 200050 [details, diff]
patch for bzr.eclass

[Comment 2 Jorge Manuel B. S. Vicetto (RETIRED) 2009-08-04 15:44:08 UTC]

Seems ok for me.

I have one question, though. Has anyone documented ${ESCM_OFFLINE}? From the use and name, I get the impression it's meant to be an SCM-agnostic variable, but a grep on /gentoo-x86/eclass only returns the use in both git and subversion eclasses, but no documentation about it.

[Comment 3 Christian Faulhammer (RETIRED) 2009-08-08 21:46:09 UTC]

(In reply to comment #2)
> Seems ok for me.

 I agree here.

> I have one question, though. Has anyone documented ${ESCM_OFFLINE}? From the
> use and name, I get the impression it's meant to be an SCM-agnostic variable,
> but a grep on /gentoo-x86/eclass only returns the use in both git and
> subversion eclasses, but no documentation about it.

 Maybe ask in -dev?

[Comment 4 Jorge Manuel B. S. Vicetto (RETIRED) 2009-08-10 15:53:43 UTC]

*** Bug 280854 has been marked as a duplicate of this bug. ***

[Comment 5 Christian Faulhammer (RETIRED) 2009-09-24 07:19:26 UTC]

Ok, there were no objections, so I just committed the patch.  Thank you very much for your patience and support.

[Comment 6 Ulrich Müller 2009-12-13 10:34:24 UTC]

It doesn't work: In spite of EBZR_OFFLINE being set to a non-empty value, it's still accessing the network during "bzr export".

This is with EBZR_FETCH_CMD left at its default value, i.e. "lightweight" checkout.

[Comment 7 Ulrich Müller 2009-12-13 16:46:10 UTC]

More findings (all tests are with the GNU Emacs repo and a 2 Mbit/s connection):

With --lightweight option:
    bzr checkout:  11 minutes
    bzr update:    < 1 minute (for unchanged repo)
    bzr export:    54 minutes (also for subsequent updates!)
    repo size:     116 MB

Without --lightweight option:
    bzr checkout:  48 minutes
    bzr update:    < 1 minute
    bzr export:    < 1 minute
    repo size:     327 MB

Looks like we shouldn't use export in the case of lightweight checkouts? Replacing it by "cp -RP -p" seems to have no adverse effects in the case of Emacs.

[Comment 8 Ulrich Müller 2009-12-13 20:20:02 UTC]

Created attachment 212915 [details, diff]
Proposed patch for bzr.eclass

Attached patch should fix the export issue and the issue from bug 296733.

[Comment 9 Ulrich Müller 2009-12-14 21:01:51 UTC]

Created attachment 213031 [details, diff]
Proposed patch for bzr.eclass

Simplified logic when to use "rsync" and when "bzr export".

I also think we don't need the einfo message about the revision, because "bzr update" outputs it already (and if we are offline the revision isn't available).

[Comment 10 Ulrich Müller 2009-12-18 07:09:18 UTC]

(In reply to comment #9)
> Created an attachment (id=213031) [details]
> Proposed patch for bzr.eclass

Committed.