Call for Discussion: New Project: Skara -- sample repacked git repo of JDK 12 sources now available

[Remi Forax]

from an island on the atlantic coast of France:

time git clone https://github.com/Project-Skara/jdk.git
Cloning into 'jdk'...
remote: Counting objects: 881253, done.
remote: Compressing objects: 100% (160363/160363), done.
remote: Total 881253 (delta 665729), reused 881253 (delta 665729), pack-reused 0
Receiving objects: 100% (881253/881253), 258.61 MiB | 1.49 MiB/s, done.
Resolving deltas: 100% (665729/665729), done.
Checking out files: 100% (66258/66258), done.

real	3m22.779s
user	1m9.630s
sys	0m12.783s

hg clone for me is 24 minutes (when there is no connection reset by peer).

The fact that github as mirrors in Europe is a huge win. 

Rémi

----- Mail original -----
> De: "Aleksey Shipilev" <shade at redhat.com>
> À: "joe darcy" <joe.darcy at oracle.com>, "discuss" <discuss at openjdk.java.net>
> Envoyé: Lundi 13 Août 2018 09:51:22
> Objet: Re: Call for Discussion: New Project: Skara -- sample repacked git repo of JDK 12 sources now available

> On 08/13/2018 07:26 AM, joe darcy wrote:
>> PPS A repacked sample git repo, as alluded to earlier in this thread and as
>> discussed in the Skara
>> presentation from the OpenJDK Committers’ Workshop, equivalent to a recent
>> snapshot of JDK 12 is
>> available for browsing and cloning from:
>> 
>>     git clone https://github.com/Project-Skara/jdk.git
>>     https://github.com/Project-Skara/jdk
>> 
> 
> Nice! Some performance data from Germany:
> 
> $ time git clone https://github.com/Project-Skara/jdk.git
> Cloning into 'jdk'...
> remote: Counting objects: 881253, done.
> remote: Compressing objects: 100% (160363/160363), done.
> remote: Total 881253 (delta 665729), reused 881253 (delta 665729), pack-reused 0
> Receiving objects: 100% (881253/881253), 258.61 MiB | 5.32 MiB/s, done.
> Resolving deltas: 100% (665729/665729), done.
> Checking out files: 100% (66258/66258), done.
> 
> real	1m25.508s
> user	1m8.522s
> sys	0m6.595s
> 
> For the comparison, pulling the .hg workspace snapshot:
> 
> $ time (curl https://builds.shipilev.net/workspaces/jdk-jdk.tar.xz | tar xJf -;
> cd jdk-jdk; hg pull;
> hg up;)
>  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                 Dload  Upload   Total   Spent    Left  Speed
> 100  752M  100  752M    0     0  5673k      0  0:02:15  0:02:15 --:--:-- 5737k
> pulling from http://hg.openjdk.java.net/jdk/jdk
> searching for changes
> adding changesets
> adding manifests
> adding file changes
> added 116 changesets with 1065 changes to 858 files
> new changesets cf34c71ca27c:802f7e5e7e6b
> (run 'hg update' to get a working copy)
> 66288 files updated, 0 files merged, 0 files removed, 0 files unresolved
> 
> 
> real	2m31.941s
> user	1m35.735s
> sys	0m22.859s
> 
> It looks that the limiting factor here is the actual network bandwidth. So, the
> performance is in
> the same ballpark, and if .hg would be smaller (my experiments saw 1.3x
> decreases with enabling some
> Mercurial options), we can get this down to 2 minutes.
> 
> -Aleksey