On Thursday 31 of May 2018 19:36:06 Slávek Banko wrote:
> On Thursday 31 of May 2018 10:21:24 gregory guy wrote:
> > Hi there!
> >
> > Sorry to step in the devs's room but I'm seeking information about
> > git uses to clone the TDE repository.
> >
> > Yesterday I applyed the method displayed in the wiki, namely:
> >
> > - git clone https://mirror.git.trinitydesktop.org/cgit/tde
> > - cd tde
> > - git submodule init -- scripts
> > - git submodule update -- scripts
> > - ./scripts/switch_all_submodules_to_head_and_clean anonymous
> >
> > After more than 6 hours of downloading (at an average of 70 kib/s)
> > the connexion stopped and wouldn't restart properly.
> > I don't think I can fetch the entire project history in those
> > conditions, this would take days...I guess!
> >
> > Is there a way to shorten the cloning in order to get the last code?
> > I tryed with "--depth=1" but I was rather unsuccessfull on the
> > matter.
> >
> > Regards
> > gregory
>
> Hi Gregory,
>
> this is strange behavior. VPS for hosting mirror.git.trinitydesktop.org
> is on a 300 Mbps bandwidth. Therefore, I do not suppose there should be
> a problem with the speed of this server.
>
> Please, can you check what the urls for the submodules are in
> tde/.git/config? Are these urls correctly replaced to
> mirror.git.trinitydesktop.org or stay referenced to the
> scm.trinitydesktop.org (master server)?
>
> Cheers

Ok, now I've tried to measure the first step with two different services 
on the same server:

$ time git clone https://mirror.git.trinitydesktop.org/cgit/tde
Cloning into 'tde'...

real    0m59.683s
user    0m27.600s
sys     0m2.452s

$ time git clone https://mirror.git.trinitydesktop.org/gitea/TDE/tde
Cloning into 'tde'...
remote: Counting objects: 60857, done.
remote: Compressing objects: 100% (22885/22885), done.
remote: Total 60857 (delta 39752), reused 56119 (delta 36438)
Receiving objects: 100% (60857/60857), 30.45 MiB | 23.63 MiB/s, done.
Resolving deltas: 100% (39752/39752), done.

real    0m5.512s
user    0m3.168s
sys     0m0.720s

This was a repeated measurement - at the first measurement times were 
doubled. The second measurements apparently benefited from the cache on 
the server. However, it is clear that Gitea gives better results while 
providing standard output during cloning.

Subsequently, I tested cloning the entire repository from Gitea:

$ time ./scripts/switch_all_submodules_to_head_and_clean anonymous

real    22m6.332s
user    10m34.756s
sys     0m53.580s

Please also try using Gitea instead of the CGit interface.
It looks like Gitea could be an advantage!

Cheers
-- 
Slávek