Re: [Arx-users] Why is arx browse so slow?

[Kevin Smith]

Walter Landry wrote:
> Kevin Smith <address@hidden> wrote:
> 
>>These are with ArX 2.2.0.
>>
>>
>>
>>Case 1 (which is just crazy and hopefully easily solved):
>>
>>---
>>address@hidden arx $ time arx browse
>>This branch does not exist in this archive
>>        address@hidden/
>>
>>real    6m18.027s
>>user    0m0.018s
>>sys     0m0.007s
>>address@hidden arx $
> 
> 
> This looks like a bug.  Can you try again with 2.2.2?

I mis-spoke. It was with 2.2.2. Does it not happen for you?

>>Case 2 (which is merely annoying and puzzling):
>>
>>---
>>address@hidden arx $ time arx browse address@hidden/arx.2
>>    2
>>
>>real    0m11.756s
>>user    0m0.068s
>>sys     0m0.016s
>>address@hidden arx $
> 
> 
> At present, ArX does not multiplex any network operations.  That is on
> the TODO list, although I think I will do it after ArX gets switched
> to a more portable networking library.

Postponing makes sense. when it lists something like 0 .. 112 is it
having to make 112 network calls? If so, then each network call is
taking something like 1/30th of a second (30 ms) which I guess sounds
reasonable.

I wonder if it really needs to descend to that level. Maybe it could
stop one level above that, unless --detail was specified. Or, if you
really feel it can be optimized away almost entirely, at some point,
without too much pain, it's probably not worth doing anything with it yet.

>From a marketing standpoint, it makes ArX appear slow. One of the first
operations someone might try is browsing a remote archive such as the
ArX tree.

Kevin