[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[lmi] VCS caching [Was: PATCH: Upgrade xmlwrapp to 0.9.0]
From: |
Greg Chicares |
Subject: |
[lmi] VCS caching [Was: PATCH: Upgrade xmlwrapp to 0.9.0] |
Date: |
Wed, 11 Apr 2018 12:46:59 +0000 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.6.0 |
On 2018-04-11 10:46, Vadim Zeitlin wrote:
> On Wed, 11 Apr 2018 01:10:20 +0000 Greg Chicares <address@hidden> wrote:
[...]
> GC> I think we should polish our usage of this:
> GC> /srv/cache_for_lmi/vcs
> GC> first. For example, if we always maintain a wx repository there
> GC> with all submodule repositories, I'm thinking we can clone it thus:
> GC> git clone --recurse-submodules --reference-if-able our-cache
>
> Sorry, I'm a bit lost, what exactly are we trying to do here?
I don't know specifically. Just "make things better".
Cloning wx from github takes quite some time:
Receiving objects: 100% (768487/768487), 301.00 MiB | 913.00 KiB/s
which is five or six minutes:
(301 / (913 / 1024)) = 338 s
Running 'install_msw.sh' here takes 13:21, i.e., 13*60+21 = 801 s.
And 801/(801-338) = 1.73, so it would be seventy-three percent faster
if I cloned wx from local cache. Well, the two or three seconds that
IIRC a local clone takes would make it only seventy-two percent faster,
but there's an enormous speedup and I want to achieve that.
> I.e. what's
> the advantage of having a wx repository in /srv/cache_for_lmi/vcs and then
> a clone in /opt/lmi/local/vcs rather than having only the latter?
I want the former, not the latter. Everything that can be cached,
should be cached in the cache directory.
> doesn't seem useful to have more than one repository on the same machine,
> what do we gain from it?
The clone in /opt/lmi/local/vcs becomes superfluous. Maybe we can
eliminate it altogether, and build in /opt/lmi/local/wx-scratch .
(I don't want to build in the cache directory, because then it's
not a pure cache directory--I'm trying to separate concerns.)
If that's no good, maybe we should still use /opt/lmi/local/vcs ,
but make it a shallow clone, or a --reference clone.
I know so little about git, and there are many options to experiment
with. Of course, if you already know the best way to achieve the
goal I'm trying to express, that spoiler would be welcome.
- [lmi] PATCH: Upgrade xmlwrapp to 0.9.0 (was: Another compiler upgrade), (continued)
- [lmi] PATCH: Upgrade xmlwrapp to 0.9.0 (was: Another compiler upgrade), Vadim Zeitlin, 2018/04/10
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Greg Chicares, 2018/04/10
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Vadim Zeitlin, 2018/04/11
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Greg Chicares, 2018/04/11
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Vadim Zeitlin, 2018/04/11
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Greg Chicares, 2018/04/11
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Vadim Zeitlin, 2018/04/11
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Greg Chicares, 2018/04/11
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Vadim Zeitlin, 2018/04/11
- Re: [lmi] PATCH: Upgrade xmlwrapp to 0.9.0, Greg Chicares, 2018/04/11
- [lmi] VCS caching [Was: PATCH: Upgrade xmlwrapp to 0.9.0],
Greg Chicares <=
- Re: [lmi] VCS caching, Vadim Zeitlin, 2018/04/11
- Re: [lmi] VCS caching, Greg Chicares, 2018/04/13
- Re: [lmi] VCS caching, Vadim Zeitlin, 2018/04/13
- Re: [lmi] VCS caching, Greg Chicares, 2018/04/13
- Re: [lmi] VCS caching, Vadim Zeitlin, 2018/04/13
- Re: [lmi] VCS caching, Greg Chicares, 2018/04/13
- Re: [lmi] VCS caching, Greg Chicares, 2018/04/13
- Re: [lmi] VCS caching, Vadim Zeitlin, 2018/04/13
- Re: [lmi] VCS caching, Greg Chicares, 2018/04/14
- Re: [lmi] VCS caching, Vadim Zeitlin, 2018/04/14