Re: [PATCHv1 0/3] git-p4: fixing --changes-block-size support

[Luke Diamand <luke@diamand.org>]

On 07/06/15 17:01, Lex Spoon wrote:
> Great work.

Thanks! I actually found the problem in my day job, so it was very handy 
having all the infrastructure already in place!
>
> For curiosity's sake, the -m solution has been observed to work on at
> least one Perforce installation. However clearly it doesn't work on
> others, so the batch ranges approach looks like it will be better.

Yes, I can easily imagine that it's changed from one version to the 
next. I tried going back to a 2014.2 server which still had the same 
problem (with maxresults), but my investigations were not very exhaustive!

>
> Based on what has been seen so far, the Perforce maxscanrows setting
> must be applying the low-level database queries that Perforce uses
> internally in its implementation. That makes the precise effect on
> external queries rather hard to predict. It likely also depends on the
> version of Perforce.

Indeed. All sorts of things can cause it to fail; I've seen it reject 
"p4 files" and "p4 print", albeit with artificially low maxscanrows and 
maxresults values. I think this means there's no way to ever make it 
reliably work for all possible sizes of depot and values of 
maxscanrows/maxresults.

Luke