Talk:Perforce

From Wikipedia, the free encyclopedia

My first wiki attempt - basicly I was reading the Revision Control entry and came across the red link so figured I would get my feet wet, I have nothing to do with the product and all information was from the company's website.

Constructive feedback appreciated User:Doctus


Good stuff - keep contributing :) As a Perforce user, I've added some comments that probably wouldn't appear on the company website. IMHO, Perforce is a dinosaur, kept alive only by skilful targetted marketing. Although it offers lots of features, the core functionality is lousy. Subversion is a far better technical system (and free). CVS probably offers the best mix of solid functionality and good support (GUIs for all platforms, off-the-shelf web-site publishing, an excellent integrated client in the altogether excellent [Eclipse|http://www.eclipse.org] IDE, etc.). User:80.194.146.18

"IMHO" is the key word there. Wikipedia is not the place for our opinions, humble or otherwise. I happen to adore Perforce, but I can't rave about it here, either. If you want to critique Perforce, the best way to do it is to find an article on a major site such as CNET that supports your claims, and link to it. Opinions -- or original research that proves Perforce inadequate -- don't belong here. Deborah-jl Talk 23:30, 17 January 2006 (UTC)
Point taken. But that goes both ways. This page reads like advertising copy, so I'm going to tone it down a little, make it neutral.
Regarding your opinion of Perforce being a dinosaur, if you can quantify how it is lousy in comparison to other SCM systems, it may be genuinely useful content for the Comparison of revision control software page. --212.159.69.172 17:54, 2 July 2006 (UTC)

Contents

[edit] Electronics Arts?

Mmm... AlienBrain is claiming that EA is one of their clients... Weird... -- lucasbfr talk 15:55, 19 December 2006 (UTC)

ok I double checked. It seems EA dropped AlienBrain for Perforce. -- lucasbfr talk 15:59, 19 December 2006 (UTC)

EA are big. They probably use more than one SCM, but they attended the summer Perforce Conference in London last year, so at least one part uses Perforce. 212.159.69.172 21:18, 25 May 2007 (UTC)

[edit] Linus, Google and Perforce

An anon editor recently added some text about Linus Torvalds telling Google engineers they should get rid of Perforce. Here's a slightly edited version:

However, Linus Torvalds, the founder of the Linux operating system which powers Google,<<ref>Anatomy of a Search Engine by Google's founders</ref>told employees of Google "get rid of it [perforce], seriously" and suggested they use Git instead.<ref>Google Tech Talk: Linus Torvalds on git, May 14, 2007</ref>

I've moved it here for discussion because Linus appears to have dissed every competing SCM system, not just Perforce. (Aside: AFAIK, Linus's criticism of other SCMs is largely because (1) they use centralized repositories and (2) they aren't good for his preferred 'non-linear' style of development via branching and merging.)

Also, does anyone know which parts of Google use Perforce? It would be nice to be more specific and less marketing-ish in the article. It would be even nicer to use a written source, not an audio. Cheers, CWC 12:09, 29 June 2007 (UTC)

Why is this even necessary to mention? He basically said that only Git is any good, and all other systems (mostly picking on centralized ones) are not worth using. This is his opinion, and last I checked opinions don't belong on Wikipedia. I watched the video when it first came out, and based on what he said it seems he never actually used Perforce or many (or most) other SCM systems that he was trashing (except for CVS and SVN, if I remember correctly). I use Perforce, and personally think it's great. I really don't care if someone else doesn't like it, even if that someone else is Linus Torvalds. His criticism of centralized SCM systems does make sense to mention in a more general SCM article (when talking about the disadvantages of centralized SCM systems, where he has some valid points), but his opinion specifically about Perforce (or any other SCM) doesn't belong on Wikipedia. Python 05:38, 16 July 2007 (UTC)
He was talking about all SCMs not designed for easy branching and merging, not just Perforce. I see no reason to mention his remarks in this article. CWC 10:20, 16 July 2007 (UTC)
Then we're in agreement. However, I have to say that Perforce does in fact have a nice and powerful branching/merging system (I use it a lot myself). Linus has apparently never used it, so his blanket comment (mainly based on his dislike of CVS, which indeed is quite bad) was baseless. However, he did make some good points about how decentralized SCM systems (such as Git) work much better for open source projects which can have hundreds of contributors from all around the world. Python 20:10, 16 July 2007 (UTC)
I am the anon user who added the entry. I think I am an excellent programmer, with a fine understanding of what makes an excellent SCM system. However, you won't find an entry for my software/company in Wikipedia. You won't find my opinions, either mentioned, or even playing a role, anywhere in Wikipedia either, because I don't think as of this writing they are "Wikipedia worthy".
So, why do you expect me to care about what you think, and just take it at face value?
Linus Torvalds on the other hand is the recognized world leader in the particular field he is addressing: source code management! For him to speak on the subject is nothing less than Ben Carson speaking about the brain.
Look at the Perforce page please:
As of December 2006, Perforce Software claims more than 200,000 developers at 4,000 organizations use the system.[1], including Google and SAP.[2] Perforce is also particularly popular in the video game industry[citation needed].
Torvalds' statement is precisely what lawyers call a rebuttal to this quote from Perforce; and it is called for in this case, because of the listed quote if nothing else. You would remove Linus Torvald's opinion, and leave this self-righteous marketing propaganda in Wikipedia, without an iota of third party references supporting it?
I have no affiliation with Linus Torvalds whatsoever (he never returns my emails, and doesn't even reply to them on public forums when they relate to him directly.) The reason I listed his opinion on Perforce, is because I noticed the Perforce page was severely lacking in objective references, and I, as a general Wikipedia user, certainly would want to know Torvalds' opinion on Perforce. And other human beings have a right to the same quality information. And your response to all of this, is that it's your opinion. Excuse me, I don't want to be harsh.
Torvalds does support, even the commercial BitKeeper SCM system, and speaks very highly of both it's creator, and the software. He does that on the same Google Video I posted as reference. Yet, the above comments, unfortunately, falsify this fact which we know is true.
I can tone down Torvalds' harsh and "unmannerly" comments all you want. I can adher to Wikipedia "good editing" conventions, if they are pointed out to me. I can live by a general, and sufficiently discussed vote, by Wikipedia editors on this subject. But I care about Wikipedia and human beings who are using it. I cannot just sit here and let this page, lacking any objective outside references, taint Wikipedia's content, without engaging editors' and other people's involvement.
I really do hope that we will resolve this amongst us. RezaRob
I saw that video with Torvalds, and I heard what he said. He didn't single out Perforce; he was voicing his dissatisfaction with the paradigm that Perforce and many other popular SCM systems currently use. Yes, he's a well known figure. And as I said before, he made valid criticisms of the centralized SCM paradigm. Those criticisms belong in the article that talks about SCM systems, not the article on Perforce.
The statement "Perforce is also particularly popular in the video game industry" may be too broad. However, Perforce's site has an article about ArenaNet's use of Perforce (along with developer quotes). [1] Perhaps this can be used as a citation for a modified version of that statement. Python 02:25, 23 August 2007 (UTC)

[edit] Unsourced?

I've removed the unsourced tags as I see three references in the 'Notes' section. If you disagree, please replace the tag and indicate here on the talk page which specific statements you feel are not adequately sourced so that this can be addressed. Thanks! Zero sharp 19:36, 29 June 2007 (UTC)

[edit] Responses to the Weakness Section

I am reviewing Perforce for my company and as such the weakness section was very interesting to me. I contacted Perforce support to get their responses to the items which I will include here because I think they are relevant to the discussion-- I think his rebuts are quite informative for these particular issues.

I also asked him previously about the "p4" command being difficult to script, and an interesting point you will see in his answers here is that they provide language bindings to their API for many scripting languages (perl, python, ruby...), meaning that using p4 in your scripts is not really what you would do if you were after fine grain control. Anyway, I played around with P4 and it does return "correct" return codes, but in some cases the whether an error occurs or not is a bit ambiguous depending on you interpretation. So its not like it always returns zero in all cases which is what I was afraid of.

Working off-line is difficult: if unable to connect to the Perforce server, a user cannot edit files not already opened for edit. The workaround is to manually change permissions on the local file to allow editing, but the user must remember to open the file for edit when reconnected to the Perforce server. Also, the user's local changes will be overwritten if the user does a sync to update their local copy, and those same file(s) were updated on the server in the interim.

It is true that in order to work on files off-line, you must change the permission locally to writable. However, once you have done work off-line and reconnected to the depot, P4V has a 'Reconcile Offline Work' feature which will visually diff your workspace against the depot to find files that have changed locally, or been added/deleted locally. This allows you to build a changelist representing your off-line work to reconcile with the depot.

If any files have changed both in your workspace and the depot, Perforce will allow you to resolve those differences (merge/accept/reject) as it would in the normal course of on-line work when both you and another user have modified the same file.

These operations are also possible from the command line; the following Knowledge Base article:

 Working Disconnected
 http://kb.perforce.com/UserTasks/WorkingDisconnected

describes the process of working disconnected generally, and how to reconcile your changes using 'p4' client commands.

I believe (again, from reviewing other calls) that what informs this perceived shortcoming of Perforce is its difference from other SCM systems, notably subversion, which use a different mechanism for enabling off-line work. Most notably, Perforce does _not_ cache any metadata locally; only revisions of files that you have synced to your workspace are available when working off-line. Particularly when compared with other revision systems, Perforce is very restrictive as it does not:

  • Allow comparison of modifications to the working base.
  • Allow files to be scheduled for addition or deletion (a slow reconciliation operation must be performed when reconnected)
  • Allow changes to be reverted.
  • Allow changed files to be sought and marked.

As with our previous conversation re: error codes, even though Perforce does allow you to work off-line, and provides methods for reconciling off-line work to insure no changes are lost, we do recognize that some users would prefer another approach to off-line work, and I will add you to the existing enhancement request for that.

Automated builds are more difficult with Perforce than with CVS or Subversion, particularly because automated creation of client specifications is awkward.

I'd have to disagree with this statement, for the following reason. The command that creates client workspaces will allow pre-created client specs to be read from the command line like so:

 p4 client -i  < my_client_spec.txt

where the file 'my_client_spec.txt' contains information about the client (most importantly the desired "View:" mapping, Submit options and the like. It wouldn't be that difficult to write a short script or BAT file to create client specs as needed, assuming you also have a programmatic way to create the spec files. Even more sophisticated scripting of client creation (and indeed almost any other Perforce command) is available via the Perforce API, for which a number of language bindings (Perl, Ruby, Python) are available.

The need to maintain branch specifications complicates merging between branches.

A branch specification is nothing more than a list of source and target mappings that is used in an integration. The 'integrate' command can either operate on file paths given in the command line, or via a branch spec. By storing often-used mappings in a branch spec, you can insure that integrations are performed the same way every time. Also, more complex mappings can be stored in a branch specification than could be done in a single command line allowing for more complicated relationships between source and targets in a branch. Of course, branching, integration and merging have the potential to be among the more complex operations in Perforce (indeed in any SCM system, I would maintain). But branch specs are (or at least are intended to be) a tool to make this process _less_ difficult.

Please don't take my word for it, though; the relevant discussion about codelines and integration are in the User's Guide:

 Chapter 6: Codelines and Branching
 06_codemgmt.html  —Preceding unsigned comment added by 15.235.153.101 (talk) 20:29, 13 November 2007 (UTC)

When backing up large sites, checkpoint creation can take considerable time during which the server is unavailable


Thank you for your effort to keep the article balanced. This is precisely the kind of argument we shouldn't be having on Wikipedia; Per the verifiability and neutrality policies, Wikipedia articles should rely on reliable sources, for criticsm, rebuttals and everything else — not the opinions of editors. I've taken the liberty of removing both the "Sterngths" and "Weaknesses" sections because they are unsourced. -- intgr [talk] 00:18, 15 November 2007 (UTC)

[edit] Licensing edits

"All Perforce software you download is fully functional, with the exception of the Perforce Server, which allows only two users and five client workspaces when used without a license. "

You can use Perforce, for free, forever, with no license. This restricts you to 2 users and 5 client workspaces. http://www.perforce.com/perforce/price.html#prices Zero sharp (talk) 06:32, 2 April 2008 (UTC)


The section title 'Licensing' has been changed to be more general and the original (_correct_) text restored Zero sharp (talk) 14:57, 2 April 2008 (UTC)