[p4] converting static labels to automatic labels

[Tetlow, Gordon]

Well, it *shouldn't* if you use the -q flag, but I haven't profiled it
enough to know. Since the server has the MD5 of each revision, that
should be enough. Of course, I suppose that diff2 isn't technically the
same as p4 flush since it's possible that there could be a file with 2
revisions that are equal. In practice, that probably doesn't make a
difference (and probably doesn't happen that much anyway).

-gordon 

> -----Original Message-----
> From: Jay Glanville [mailto:Jay.Glanville at naturalconvergence.com] 
> Sent: Friday, July 20, 2007 10:45 AM
> To: Tetlow, Gordon
> Cc: Perforce Users Mailing List
> Subject: RE: [p4] converting static labels to automatic labels
> 
> Gordon, interesting suggestion.  Thanks.
> 
> But, isn't diff2 actually doing file content comparisons?  Isn't that
> going to be a lot more costly for the server then simply 
> doing revision
> number comparisons?
> 
> Also, doesn't that fail when my label's view is scoped down?
> 
> For example, if the view of my label was
>   //depot/path/a/...
>   //depot/path/b/...
> But the label didn't include
>   //depot/path/c/
> 
> Then, when executing:
>   p4 diff2 -q //... at label //... at changelist
> It will include //depot/path/c in the diff, giving output like:
>   ==== <none> - //depot/path/c/filename.txt#5 ====
> Ie: not (<none>) in label, but in change list.
> 
> 
> Thanks for the suggestion.
> 
> JDG
> 
> 
> 
> > I would go a step further and not use a workspace at all but diff2:
> > 
> > p4 changes -m 1 //... at label
> > p4 diff2 -q //... at label //... at changelist
> > 
> > That should be faster because it doesn't have the overhead of the
> > metadata operations a p4 flush does.
> > 
> > -gordon 
> > 
> > > -----Original Message-----
> > > From: perforce-user-bounces at perforce.com 
> > > [mailto:perforce-user-bounces at perforce.com] On Behalf Of 
> > Stephen Vance
> > > Sent: Friday, July 20, 2007 7:15 AM
> > > To: Jay Glanville
> > > Cc: Perforce Users Mailing List
> > > Subject: Re: [p4] converting static labels to automatic labels
> > > 
> > > There are a couple of things to consider. Which ones you 
> worry about
> > > depends on your trust in the orderliness of your environment.
> > > 
> > > The easy part is
> > > 
> > > p4 changes -m 1 @label
> > > 
> > > which, as you note, answers the question "What is the highest 
> > > changelist
> > > number that contributes to the label?"
> > > 
> > > Next, if you create a workspace and
> > > 
> > > p4 sync -k @label
> > > p4 sync -k -n @changelist
> > > 
> > > you will find out if the label purely represents the 
> > > changelist *within
> > > the constraints of your views*.
> > > 
> > > However, both your workspace view and your label view can 
> > potentially
> > > split changelists. If you open up those views to include 
> your entire
> > > repository, you could use a combination of 'p4 sync -k -n 
> > @changelist'
> > > and 'p4 labelsync -n -l label @changelist' to easily 
> > > determine if it was
> > > truly pure. If you need to keep the views restrictive then 
> > it's a more
> > > computational task to determine.
> > > 
> > > Steve
> > > 
> > > Jay Glanville wrote:
> > > > Hello all.
> > > >
> > > > I'm wanting to convert some of the old static labels we 
> have (many
> > > > containing over 60k files) into automatic ones, in the 
> > hopes that it
> > > > will make label diffs more efficient.  The conversion 
> > process itself
> > > > isn't that hard.
> > > >
> > > > The hard bit is trying to determine if an automatic label 
> > > can represent
> > > > a static label.  The only way I can think of right now is 
> > > to find the
> > > > highest change list in the label, and then perform a diff 
> > > between the
> > > > label and the change list.  But, is that the best way?  
> Is there a
> > > > simpler way (that just hasn't occurred to me yet...).
> > > >
> > > >
> > > > So, my question to these eminent minds is this: what is the 
> > > best way to
> > > > determine if a static label can represented by an 
> automatic label?
> 
> 
> ---
> Jay Dickon Glanville
>