Re: git on small systems

[Matt Sporleder <msporleder%gmail.com@localhost>]

> On Feb 9, 2017, at 4:03 AM, Roy Marples <roy%marples.name@localhost> wrote:
> 
>> On 09/02/2017 02:38, matthew sporleder wrote:
>> Sorry to revive this old thread but after reading some news about
>> Microsoft and other big players moving *GIANT* repos to git (always
>> with special considerations) I thought it would be worth testing our
>> own larger repo.
>> 
>> Without any of the special tuning  (well I still use threads = 1) and
>> the latest git  (2.11.1) I am able to clone full src with a peak
>> between 350 - 400MB memory usage over ssh, which is higher memory than
>> http.
>> 
>> This does appear to be a moderate improvement over our initial testing
>> two years ago.
> 
> Why would you want full src on such a low memory device?
> git already has shallow clones - and since git 2.0 at least you can push
> from them.
> 
> Can you measure a shallow clone --depth 1 clone please?
> 
> Roy

I did in 2015. Even the rpi could handle depth=1 just fine. 

This was all because cvs can operate in very low memory environments.