The Brad Hole

Ugh - Perforce. Do you still have to tell it you want to edit a file, instead of just editing it and then telling it later? I hated that.

We use Subversion at work, but instead of just hosting thru Apache, we most often connect thru a ssh-tunnel with the "svn+ssh" protocol. It's much faster and works like a charm.

On the other hand, it requires a ssh-account for everyone accessing and that you either change the umask for those connections or that you set the "set-user-ID-on-execution" or "set-group-ID-on-execution" on the svnserve util.

I find offline editing of PerForce files pretty easy - just change permissions, then, when back online, simply do "p4 edit" on the changed files (or p4 diff -x to find the changed files and update depot accordingly) and you're in business. If a file has had an edit while you were offline, you'll have to do a p4 resolve, but that's nothing new. Basically, offline editing in p4 simply swaps the "p4 edit" and actual file editing steps - you do "p4 edit" *after* editing the files.

I have no experience with SVN, but I've been very, very happy with p4. One question though - does p4v still only work with mac clients (as opposed to the darwin client - we use the latter at Luxology since our modo app is X only and doesn't have resource forks).

Regarding the "Perforce makes me open a file before I can edit it" issue: All commercial version control systems work that way. If you are working on project with other people, the system needs to know when a file is "checked-out". That way, it can warn other users that they may need to resolve their changes if they try to edit the same file at the same time.

Perforce will also list your open files as part of a pending changelist, so that other users can see what changes your are working on.

Of course, if you're the only person working on a given project, you can do whatever you want.

"I find offline editing of PerForce files pretty easy - just change permissions, then, when back online, simply do "p4 edit" on the changed files (or p4 diff -x to find the changed files and update depot accordingly) and you're in business."

Thanks for the tip Matt. This still leaves me with a problem though. Under 10.4, my p4 client workspace won't be able to connect to the p4 server for the near future, so I can't run a p4 diff command to determine what I've changed. I guess I can copy my full tree over to a connected Mac and do a diff using FileMerge or CodeWarrior though.

As for p4v and Darwin, I assume it still doesn't work. The Mac p4v app still uses an Aqua version of Qt. They'd probably need a separate Darwin app that uses an X11 version of Qt I'd guess. They're pretty responsive to support e-mails - I'd drop them a line to see if they have plans.

Ah, yes Perforce... After having to suffer with VSS for 5 years we finally switched to Perforce for 99% of our company projects (ironically the 1% that hasn't migrated over is 50% of my work...). Any flaws I've noticed with Perforce are quickly outweighed by performance.

Brad, do you still have a 10.3.9 partition? If so, you could reboot into it and do work there. Or do you absolutely need to do work in 10.4.0?

"Brad, do you still have a 10.3.9 partition? If so, you could reboot into it and do work there. Or do you absolutely need to do work in 10.4.0?"

I luckily kept my G5 running 10.3.9, so I didn't totally screw myself over. I also installed a 10.3.9 partition on this g4, but that has some gotchas. First, I have to use a fresh p4 workspace - I can't recycle my old one without royally screwing things up. Second, MovableType is running under 10.4, but isn't set up for 10.3.9, so I'd have protracted outages for that and ftp services.

But I need to debug KotOR's crash on 10.4.0 and nVidia cards. I also had to test netplay in Tiger Woods across my 2 Macs. I got into a rhythm where I'd do all my code changes on the G5, then just copy over the binary to my 10.4 box and run it straight, no debugger. This worked well for working on the net code in Tiger Woods, but it's not really feasible for KotOR. I may have to end up doing remote debugging from the G5 to the G4 to work on KotOR.

New version of the Cisco VPN client:

http://www.versiontracker.com/dyn/moreinfo/macosx/12696

We've never really had a problem with 2 people editing the same file before. Although sometimes it's possible that 2 people might touch different parts of a file, we usually know what each other are doing because we see who's assigned bugs to themselves (so if I see someone assign an audio bug to themselves, then I know that I should talk with them before touching the audio code).

When a conflict occurs, then the second person to check their code in just does a merge.

Maybe this isn't the case in big organisations, but if 2 people are rewriting the same code at the same time then isn't that indicative of a deeper problem with communications?

"Maybe this isn't the case in big organisations, but if 2 people are rewriting the same code at the same time then isn't that indicative of a deeper problem with communications?"

Not necessarily. We have a porting library here that is shared with most of our game ports. It's very common for several ongoing ports in development to use this library, and for 2 ports to require a change in the same file at roughly the same time as we flesh things out more and more. For example, the Sims2 team might be implementing some wacky DX9 feature and the Tiger2k5 team will need to do DX9 work as well. We don't set up bug lists until the games hit beta, so during early development, things are a bit more free-form.

With p4, we can easily see if someone else is tinkering with a file and find out if they need something I'm going to work on or vice-versa. In this regard, p4's locking mechanism actually helps communications.

any word on the apache bug with mac os x 10.4? is there an apache bug number? i'm dying for a fix - any fix.

I liked subversion until I tried arch:

http://gnuarch.org/

bit of time to learn but it really pays off in elegance.

Michael

By the way, if $FILE is a whole file path, $FILE/rsrc is the path to the resource fork... Couldn't you use that for Perforce?

I've been using Perforce at my new position for a little over a month now. Before that, I used Subversion, and prior to that Oracle SCM, CVS and VSS. Before Subversion, Perforce would have been worth the hefty sticker price, but with Subversion offering atomic commits and a much simpler, more intuitive command model, I don't see any justification for companies to continue to hemorrage cash paying for Perforce. As for performance, it may just be my company network, but its performance is lousy compared to Subversion. P4Win is klunky, unintuitive, and goes "white-screen" during each sync, which is very slow. Other services on our network respond reasonably well, so I'm inclined to believe that the sluggishness is with the Perforce client. Subversion still has some maturing to do, but at a sticker price of $0, It beats Perforce hands-down. Subversion does have a UI, however. TortoiseSVN is a much better UI than P4Win, and it is integrates with Windows and Linux explorers.

"P4Win is klunky, unintuitive, and goes "white-screen" during each sync, which is very slow. Other services on our network respond reasonably well, so I'm inclined to believe that the sluggishness is with the Perforce client. Subversion still has some maturing to do, but at a sticker price of $0, It beats Perforce hands-down. Subversion does have a UI, however. TortoiseSVN is a much better UI than P4Win, and it is integrates with Windows and Linux explorers."

I believe P4Win is deprecated in favor of p4v, which (IMHO) is excellent. I can't say I've had a performance problem with p4 (I access ours remotely through a vpn) so I'm inclined to think that P4Win vs. p4v is part of your problem. Beyond that, my interest in comparing p4 and svn is strictly limited to the Mac. I couldn't care less about TortoiseSVN or P4Win.

Anyone have a performance problem through a hardware Cisco VPN firewall using Cisco VPN client 4.0.3(F) using Perforce server 2005.2 ?

-- Or any VPN performance issues that are of note?

"Anyone have a performance problem through a hardware Cisco VPN firewall using Cisco VPN client 4.0.3(F) using Perforce server 2005.2 ?"

We don't use that exact setup (I use a later Cisco client with an earlier p4 server), but I'd say performance is pretty good. I used to access a p4 server directly (no vpn) but I don't recall if that was faster. I don't remember it being different.