Talk:IBM DevOps Code ClearCase

---

Changesets

This part seems wrong:

> Transactions not atomic: Changes to files or directories are independent from others, unlike some other systems where multiple changes can be committed atomically at the same time, using the concept known as Changeset. Atomicity is desirable because it allows multiple changes that address a particular problem to be committed as a single unit, and, provided the developer has been careful, results in a buildable code base for every version.[6]

I distinctly remember working with change-sets in ClearCase. —Preceding unsigned comment added by 68.7.99.254 (talk) 07:45, 2 December 2007 (UTC)

Sure - it's easy to cite:

http://publib.boulder.ibm.com/infocenter/cchelp/v7r0m0/index.jsp?topic=/com.ibm.rational.clearcase.hlp.doc/cc_main/t_view_changeset_act.htm

Note that the wikilink doesn't really point to a discussion of "ChangeSets", but to "Atomic Commit" - something different. However, look back to the history where that detail was introduced apparently by the IP-editor who's been hacking at the article recently - there's been other debris removed from that edit. Tedickey 20:21, 2 December 2007 (UTC)

Transaction are NOT atomic. Sure - you a have a change-set of a commit (called delivery), but it is not atomic. If the process stops or is being cancelled midway, no rollback is being performed. That's the definition of atomic in my book. OmerMor (talk) 17:07, 27 July 2009 (UTC)

pro/cons

no comparaison against other SCM ? This page looks too much like ClearCase marketing... No mention of its *extremely* slow speed, unfriendly user interface and terminology?

terminology??? Ever used an RDBMS at all? --Afc 03:15, 28 July 2006 (UTC)

The friendliness of a user interface is subjective. I like most of it. Jgrahn 19:54, 13 April 2006 (UTC)

It's very well designed and not at all unfriendly, but indeed it requires resources in terms of computing power and performant io disks -- mlz

No dev can argue that it is "well designed" and "user friendly", that is simply false. I enjoy its positive sides (stability, support for large development organizations) yet it is slow due to its old architecture and suffers severely bad usability (can you say "modal dialogue layering"?) - I have never found any usability engineer or any dev with competence in usability argue otherwise, simply because they break a host of usability principles. Again, the product has positives. - Marcus Widerberg / senior dev

—Preceding unsigned comment added by 192.71.68.101 (talk • contribs) 20 June 2006

The Senior Dev guy is probably talking about the Windows GUI, and I tend to agree there. I was (as User:Jgrahn) talking about the Unix command-line, version-extended pathnames, manpages and so on. JöG (talk) 07:59, 16 February 2008 (UTC)

How can I know the all the files checked in by an induvidial

How can I know the all the files checked in by an induvidial?

If you're looking for the specific ClearCase command, it is: cleartool find -avobs -version "created_by(username)" -print (read the find and query_language manuals for details) Trent 15:40, 28 March 2007 (UTC)

Reputation of reliability... hum that's not that I saw with 3 different clients. It is used in big project (because from IBM) but it's not more reliable for that.

--

The same here. Entire days of downtime for server problems. Because the files exist only on the dynamic view, it is not even possible to work on the local copies of the files. But it is from IBM so it must be good. 83.145.211.38 (talk) 04:04, 12 December 2007 (UTC)

Integration

At some point, I would like someone to explain to me exactly how ClearCase integrates well with Rational Rose, ClearQuest and other IBM/Rational tools. In my experience, those other tools resist any kind of serious revision control. Jgrahn 19:58, 13 April 2006 (UTC)

Jgrahn, Rose integrates quite well with ClearCase, you can invoke most of the ClearCase functionality from its integration menus. I cannot believe you're talking seriously about ClearQuest integrations, since it complements CC functionality, and it's data is not really meant to go through version control. If you were ever near Rational Software Architect (or plain vanilla Eclipse, btw) you can see some other nice example of integration with ClearCase.

Afc 23:07, 18 March 2007 (UTC)

It's been a long time since I wrote that, but concerning Rose: IMHO to "integrate" with ClearCase, you need to do more than adding a menu items with invokes "cleartool co" etc. Solving (not just claiming to solve) the merge problem would be on the top of my list. JöG 21:44, 6 November 2007 (UTC)

Neutrality

Rather than a "criticism" section (ClearCase is not a person, as much as I'd like to anthopomorphize it), I think we should definitely add a pros/cons section that references the SCM tools comparison. --Afc 03:15, 28 July 2006 (UTC), a very biased source.

As others have said here, this article reads like marketing blurb for the product. I think it should contain at least a section of "criticisms" -- I am a professional software engineer and have heard many. Alf Boggis 12:03, 31 May 2006 (BST)

As someone new to ClearCase and the article, I have to say the article reads more like a "ClearCase is SCM and this is what makes it unique". I do believe this is article is more NPOV than positive-POV, and it helped me understand why my company has decided to use it. However, before I erected a ClearCase service I'd prefer someone make a Criticisms section (even though CC is too expensive for me to buy just because of a Wikipedia article). --Caidence 15:28, 12 June 2006 (UTC)

I added links to "List of revision control software" and "Comparison of revision control software".--Btwied 17:01, 14 June 2006 (UTC)

A criticism section would be great, I also have heard many criticisms of ClearCase. There must be some published articles out there that could be referenced. --JRavn 23:02, 24 July 2006 (UTC)

I should clarify that the criticisms I have heard are with the UCM and the Windows GUI parts rather than basic ClearCase Alf Boggis 14:11, 27 July 2006 (UTC)

I could certainly write up a criticisms section, as I have been working with ClearCase (as an admin) since 1994 and am very familiar with its flaws (and advantages, of course). But wouldn't that count as original research? Maybe I'll finish up the critique page on my own website first. Trent 15:36, 28 March 2007 (UTC)

Funny how the criticism section disappeared recently (considering it was pretty mild) Tabgal 00:05, 15 October 2007 (UTC)

Contributions show only the one change by the IP that did the edit. Presumably someone will add it back (preferably someone who's actually familiar with the product ;-) Tedickey 00:18, 15 October 2007 (UTC)

I have issue with this statement:

Unix/Windows Interoperability: VOBs hosted on *nix (Solaris, Linux, AIX, HP-UX, IRIX primarily) servers can be accessed from views hosted on Windows clients. VOBs hosted on Windows servers can only be accessed by Unix clients with snapshot views.

Isn't that statement a con--that you can't use the same access methods in multiple operating systems? 128.244.70.240 (talk) 11:20, 16 April 2008 (UTC)

It seems to be only commenting that (while the product provides some degree of interoperability) there is a limitation of the Windows server implementation. The second sentence could be (assuming it's reliably sourced...) moved to weaknesses. Tedickey (talk) 11:28, 16 April 2008 (UTC)

Are references to third party patches acceptable (disclamer I wrote it)? The reason dynamic views on a *nix client fail when the VOB is hosted on Windows, is because windows returns the path backslash separated instead of the *nix forward slash, that's it. Is it appropriate to say why the dynamic views fail?

Unix/Windows Interoperability: VOBs hosted on *nix (Solaris, Linux, AIX, HP-UX, IRIX primarily) servers can be accessed with dynamic views, snapshot views, or the new web protocol based client: the CCRC on Windows clients. VOBs hosted on Windows servers can be accessed with snapshot views or CCRC from Unix clients, but not dynamic views due to the Windows server returning file strings with backslashes as the path delimiter. There is a 3rd party patch for Linux to work around that issue.

Dfries (talk) 00:56, 26 July 2009 (UTC)

Weaknesses section has weasel words

This section is poorly written, e.g. "unlike recent SCM, you cannot commit a bunch of files" and "they accumulated a ton of code and are based on old architecture that make them slower and difficult to use that they could". All the weaknesses are rather subjective, and don't have any citations. Hertzsprung 17:21, 29 August 2007 (UTC)

It's not just weasel words, it's absurd. Who can tell how much legacy code ClearCase contains? Who waits for minutes for "the simplest changes"? What version control tools (except RCS and SourceSafe) cannot easily be used to generate "excessively complex" systems of information? Who cares if transactions are atomic, when she is alone on her own branch? Et cetera. JöG 21:56, 6 November 2007 (UTC)

I care about atomicity, even when I'm alone on my own branch. Ever had a network failure in the middle of a commit? —Preceding unsigned comment added by 62.119.149.232 (talk) 09:08, 6 December 2007 (UTC)

What exactly counts as a credible citation? I don't believe that any neutral scientific papers would publish criticisms of consumer products and the management-oriented magazines are not going to bite the hand that feeds them. I could make a benchmark that shows how fast (not!) Clearcase is on a reasonable corporate network, but would anybody believe it? By the way, I can't believe that the facts that the dynamic just disappear when connection to server is lost and that Clearcase running dynamic views does not survive changing IP address (DHCP and laptops, anybody?) are not mentioned. No I agree that the Clearcase Explorer looks pretty fine until one attempts to use it and the Version Tree tool is actually good, but the drawbacks make Clearcase unusable. Now of course one can work around bad tools. I would rather work on hard problems using good tools. 83.145.211.38 (talk) 04:04, 12 December 2007 (UTC)

I used CC from '98 to '03 or there abouts. I had three complaints

• it was expensive.
• dynamic views needed serious network capability and really good work habits. I just used snapshots and an out-of-band update notification mechanism
• its multisite support was primitive and painful to use. Away from The Server you basically lost all revision control, even with snapshots

Don't know what's like now - current work environment uses something positive neolithic. —Preceding unsigned comment added by Thoglette (talk • contribs) 14:48, 1 September 2008 (UTC)

Weaknesses Citation is from a competitor

I have an issue with citation 12, which apparently justifies this paragraph :

"Speed: ClearCase dynamic views are slower than local filesystems, even with good network infrastructure. A benchmark shows that many basic tasks take about twice as long to run using a ClearCase dynamic view than on competing SCM tools. Basic operations on snapshot views can take more than ten times as long as on competing systems. Although initial builds using dynamic views require roughly twice as much time versus using a local filesystem [12],"

If you look at the source in 12 it is made by a competitor. Having used clearcase I know it is slow, however if I was a new user I would not trust a competitor's marketing data as the best information. -mike

Yes, I wrote that piece and I know. If you know of an independent benchmark, please cite it. The problem is, you can not just make benchmarks yourself and publish them because that would count as original research. Usually version control benchmarks do not compare build times because the entire idea that accessing files in source control would be slower is so absurd.

In my experience, those numbers are rather biased to favor ClearCase. Where I work, compiling on dynamic view takes about 2.5 times as long as on a plain file system or CC snapshot view. And that's on a slow encrypted hard disk. I would not believe it myself if I didn't have first hand experience. 192.100.130.7 (talk) 12:24, 18 January 2008 (UTC)

I added a reference and some information concerning speed-performance for windows clients. Erapade (talk) 22:17, 30 January 2008 (UTC)

I agree that the speed issue is worse than even Perforce was willing to claim. We found a minimum 2.5x slowdown on everything, even with a good network. On a slow network, it was completely unusable, even with snapshot views--over our in-house (and not too bad) intercontinental WAN connection, it would take days to check out any sort of complicated snapshot view. Even a trivial one took hours. —Preceding unsigned comment added by 217.10.60.85 (talk) 13:46, 11 March 2008 (UTC)

I think most critism here is due to lack of knowledge of the product. I have been a user of ClearCase since 1997. I used both snapshot and dynamic views and set it up in two companies without being an administrator. The claim that creating a snapshot takes a long time is, I'm sorry to say, BS. For dynamic views, yes, your network may need to be tuned a little. If you have a good admin, it is no problem, he can find the network bottlenecks. I have met a lot of people that critize CC without understanding the basic branch-work-merge paradigm. To me anyone who thinks branshes are unneccessary don't understand version control. I have seen haters of ClearCase that insist on modifying the same file by multiple people at the same time. This goes against the concept of a "view" and a "branch". All I know is that the used all features of this tool including the "multi-site" aspect between France and USA. It works. You may not know how to configure and use it. That is your fault not the product's fault. - Oguz —Preceding unsigned comment added by 216.255.117.130 (talk) 19:28, 20 March 2009 (UTC)

More history

More information about Clearcase versions and corresponding dates is missing in my opinion. Does anyone have some information? It's difficult to find some on the Internet. 84.152.207.18 (talk) 08:34, 25 July 2008 (UTC)

Nothing unique about majority of features

I changed a title from "Unique Features" to "Features", because clearly there is nothing unique in at least majority of the features listed. These are VOB, Interoperability, Integration. It may not be called VOB, but other SCM systems also maintain the data present in the VOB. 202.80.48.175 (talk) 11:00, 15 January 2009 (UTC)

It's not brought out clearly in that section, but go up to "Views". What was novel about the VOB was that it was implemented as a virtual filesystem. DSEE was implemented using one of Apollo's features, the ability to have customized filesystems (an example would be compressed files, encrypted, etc. - the Apollo feature dated from the mid-1980s). When ClearCase was implemented, it made this available in a more/less portable implementation, not relying on the Apollo features. Tedickey (talk) 12:07, 15 January 2009 (UTC)

Weaknesses

I've been using clearcase for over a decade and I can say that this section is clearly BS. Clearcase is a large expensive product, which requires loads of resources/bandwidth to use correctly. It reads as though it's been written by a competitor, Perforce and uses weasel words. It begs the question, why would somebody put a weaknesses section in. Taking each point in turn.

• Transactions not Atomic. That is BS in its purest form. It would invalidate the whole concept of CM if this was the case, and is using its standard operation as a weakness.
• Aging. "have stagnated and accumulated significant quantities of legacy code". So what, all apps have legacy code, that is one of the fundamentals of software development, a common concept, but not a weakness. "ClearCase was designed to support existing build systems rather than require the developer to use only ClearCase tools such as clearmake." A design choice isn't a weakness. The fact you could use your own build tools, and optimize the CM process for your own shop was a selling point.
• Speed. "dynamic views are slower than local filesystems" That was probably the case in older versions, but generated view for any app, across a network, unless your using 100Mbit or 1Gbit ethernet will be slower. Its a high end product.
• Speed - Windows clients. "meaning this is a historic argument.". Clearly a case for History of Rational ClearCase, not a salient view of the current product.
• Sensitivity to network problems. The catchall. Show me an app which operates online, which is now the majority of them, which is not sensitive to network problems. That is a completely redundant bullet point.

If the sections it really needs to be radically overhauled. Personally I think it should be removed. scope_creep (talk) 17:46, 1 April 2009 (UTC)

The section should be removed as it offers no real value. scope_creep (talk) 17:46, 1 April 2009 (UTC)

I have been using clearcase for over a decade too, and the weaknesses section is too mild in my opinion:

• Speed. It is extremely slow, even in snapshot views (every single operation takes time because it has to go over the network). Compared to svn or git which are also free is has no single advantage. SCM should be fast enough to be unnoticeable by the developers. CC is extremely noticeable.
• Maintenance and Reliability. This tool is the epitome of high-maintenance. It breaks on every occasion. You must have a full-time CC administrator on-site to make it work.
• Price. Very expensive tool, and the free tools are not only free but also better in every aspect.

I am subjective of course - but I also talk from lots of (painful) experience. OmerMor (talk) 17:24, 27 July 2009 (UTC)

You can't discuss price without including a citation that compares it to other similar products. To say it is expensive compared to free opensource products is not appropriate. SVN should also be added as "Subversion" with a link.

Maintenance is also factually incorrect and subjective. We operate the largest ClearCase implementation in europe with only 2 administrators. It would be better to say that the product is complex and requires 3-5 days classroom training in order to perform administrative duties. You can cite IBM Rational's Training bit.ly/CU9Pv Xantiriad (talk) 10:07, 30 July 2009 (UTC)