copy folders instead of sequential hg clones

Dmitry Samersoff dmitry.samersoff at
Mon Apr 14 16:18:43 UTC 2014


After local clone your .hg/hgrc file contains

default = <your location>

instead of

You have to fix it to do update/push. And it's the only thing you have
to fix.


On 2014-04-14 20:13, Pete Brunet wrote:
> On 4/12/14 4:32 AM, Dmitry Samersoff wrote:
>> Peter,
>> It's perfectly fine to just copy entire forest.
>> Local clone is faster than copy, but you have to restore .hg/hgrc
>> afterward if you plan to do the push from this workspace.
> Hi Dimitri, I tried local clone first and that was the first thing I
> noticed so I was concerned there might be other problems and decided to
> cp my directory instead.  Is the .hg/hgrc the only file I'd need to copy
> after local cloning?
>> My typical workflow:
>> 1. The script mirrors hs-rt and dev to dedicated machine every night. It
>> is always a full clone from scratch.
>> 3. Starting a new CR, I create a directory with CR number, file called
>> "comments" with ojdk push message and copy nightly workspace to it.
>> 4. When fix is ready, I create a webrev against local ws (webrev -N) and
>> put it to using sshfs or special script.
>> 5. After all reviews, I go to one of shared server, clone fresh
>> workspace, apply patch form webrev to it and submit a commit job.
> Thanks for that.
>> -Dmitry
>> On 2014-04-12 05:12, Pete Brunet wrote:
>>> Hi Jon, I am on VPN from Austin.  It has always taken forever.  And
>>> downloading programs from the internet when on VPN takes forever.  I
>>> started looking into this a while back and I don't know if this is the
>>> issue but I found that the DNS servers are in Europe when I am on VPN. 
>>> Right now as reported by ipconfig /all my Cisco AnyConnect DNS servers are:
>>> in Madrid
>>> in Amsterdam
>>> Pete
>>> On 4/11/14 7:07 PM, Jonathan Gibbons wrote:
>>>> To me, this message raises the questions of "why does it take
>>>> forever?" and, "how long should it take?"
>>>> I see a wide range of expectations. Some folk can clone fast, and
>>>> assume that everyone else can as well.  And then there's reports like
>>>> this, that it takes "forever".
>>>> I'm all in favor of doing whatever it takes to speed up the workflow,
>>>> but where is the threshold between "yeah, it really does take this
>>>> long" and "if it takes this long, something must be broken"?
>>>> -- Jon
>>>> On 04/11/2014 04:26 PM, Pete Brunet wrote:
>>>>> Since it takes forever to clone on my Win machine, in the case where I
>>>>> want to work on several bugs, is it OK to instead clone the first
>>>>> directory and then cp -ar that to n additional directories?  -Pete

Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.

More information about the build-dev mailing list