Talk:Sun WorkShop TeamWare

Sun external link

The external link no longer works. -- Bartosz 23:02, 1 June 2006 (UTC)

Thanks for pointing that out. I've replaced it with a link to a live page that also discusses TeamWare. -- Seitz 05:15, 2 June 2006 (UTC)

Rename to avoid confusion with Teamware Group Oy

This article is meant to discuss the TeamWare revision control software published by Sun Microsystems. Another editor pointed out that TEAMWARE is a trademark for software products produced by Teamware Group Oy. To avoid confusion and keep this article focused on the Sun revision control software, I would like to rename this article to "Sun TeamWare". Any objections?

Perhaps "TeamWare (Sun Microsystems)", or "Sun WorkShop TeamWare"? -- Seitz 22:51, 20 October 2006 (UTC)

I'm moving to Sun Workshop TeamWare. TeamWare et al will still link here.--NapoliRoma 20:30, 4 June 2007 (UTC)

Credits

An editor recently removed any mention of Larry McVoy, claiming personal knowledge that McVoy was never a member of the TeamWare team. This contradicts claims made on McVoy's company's web site. For now, I've restored McVoy's claim as "the designer of TeamWare", with a reference to the BitKeeper site as a source. I've also kept the claim that Evan Adams was the lead architect, and added a reference to a paper where Evan Adam discusses developing TeamWare.

I've started to do some more research on this issue, and it appears there is some controversy over McVoy's role in TeamWare. I will try to follow up when time allows. More information with references would be welcome. Seitz (talk) 14:39, 19 March 2008 (UTC)

I recently had some mail exchange with Glenn Skinner and he did not mention Larry as inventor. As I plan to implement smoosh for SCCS, I may then be able to compare the mlisp implementation from Glenn and the 1998 c Implementation in Bitkeeper with the patent paper. Schily (talk) 13:38, 27 February 2015 (UTC)

History

I am writing an thesis about DVCS and I am missing the date of origin of TeamWare in this article.

Please have a look at the article about [Distributed revision control] and [Code Co-op] then compare the assertion that Code Co-op is the first DVCS to the statements you can find here about the age of TeamWare: OpenSolaris SCM Project History, Source-code management for an open JDK and Interview with Larry McVoy.

As already stated in [Talk:Code Co-op] TeamWare seems to be much older than Code Co-op and therefore should be considered as the first DVCS if any. The least that could be mentioned is that TeamWare's history dates back into the 1980's.

(Sorry for not logging in, but my account on de.wikipedia.org does not seem to work here. And by the way I am not a native English speaker.)

78.42.205.197 (talk) 21:23, 17 April 2009 (UTC)

Thank you for the interesting pointers, you could send me a mail if you like.... BTW: NSE (Network Software Environment) existed before TeamWare. It was made available by Sun in Spring 1986 with the Sun3 Hardware line. NSE was abandoned because HP and IBM did not like to implement the translucent filesystem that was needed for NSE. Note that TeamWare was created in the late 1980s because Sun HP. IBM and others liked to collaborate for the CDE development. NSE did not implement bringover like TeamWare, but it allowed people to NFS+TFS mount the master tree and do concurrent development. NSE already allowed collaboration but need a special high level SCCS filesystem check. It seems that it was not atomic. Is there anybody who is able to supply me with NSE documentation? Schily (talk) 14:00, 27 February 2015 (UTC)

smoosh?

Larry McVoy sent mail to the cssc mailing list in July 1998 mentioning smoosh, and making it clear that his implementation for BitSCCS did not use Sun's patented algorithm. Quoting from the mail

By the way, smoosh is that all likely to work right now, I stuck it in there

just to see if I could - it was all of about 20 minutes of work. Sun has the original smoosh alg patented (it's a very cute one pass deal that is quite fast) so I couldn't reuse that idea. So the smoosh that is there just does the brute force: apply the deltas one at a time approach. I think that will be fine because (a) this version of SCCS is pretty darn fast, and (b) smoosh is a rare event as a percentage of the files processed (around 1% or so). So I was optimizing for the uncommon case back at Sun. Live and learn.

Since a patent would have required an implementation, that leads to the conclusion that he was not the original author of smoosh. (I have a copy of BitSCCS-0.5.1-SCCS, from which "20 minutes of work" seems plausible) TEDickey (talk) 01:54, 27 February 2015 (UTC)

BitSCCS's version was distributed as SCCS files, begun in April 1998. An older version (different implementation) is alluded to in an informal paper by McVoy in February 1993 (see ftp://ftp.cs.uni-sb.de/pub/sun/papers/smoosh.ps.gz -- but trim the leading junk from the file). Its acknowledgements section clarifies that smoosh was a collaboration between Skinner and McVoy: Skinner developed a prototype). Quoting:

Glenn Skinner prototyped some elisp code that showed that his ‘‘after the fact’’ line of thinking, the basis for smoosh, would really work. I took that idea and prototyped some stuff in perl and finally gave up and made the current C version. I’m responsible for the one pass algorithm, unfortunately, since it is the buggiest part of this code. Bill Shannon encouraged me to go off and hack on this and has hashed it through many times. Glenn Skinner has hashed this through even more countless times and much of what I’ve learned about SCCS I learned from him (the rest came from looking at the damn s-files, the documentation talks about what not how).

TEDickey (talk) 09:44, 27 February 2015 (UTC)

If you have a working version of smoosh from Bitkeeper, it would be interesting to check.... I believe I fetched the latest OSS version of BitKeeper 0.5.2 and it has a smoosh that claims to work. The most recent version (when called as documented) does effectively nothing even though the code creates the assumption it could.

The mlisp code Glenn send me a few weeks ago is much longer and creates the impression to work (I did not try to check thus actually).

Checking whether or not the implementation from Larry uses the patented algorithm would take some time. The mlisp code from Glenn is 1640 lines of code, the c source from Larry is 240 lines. The mlisp code calls get(1) and delta(1), the c code from Larry calls library versions of get and delta, so I see no real difference here.

Anyway, SCCS is so fast that it doesn't matter if you have to call get(1)/delta(1) to apply deltas to an existing history file if there is no method to directly create the new interleaved delta. My library version of get(1) extracts an average source file (10kB extracted size) in 190 microseconds and if you read the delta table once and just repeat the interleaved delta extraction for the various revisions, each iteration takes less than 100 microseconds (in other words, you can go through the history of a file with a time slider and produce annotated versions of a file with an update rate of more than 10000 per second). Merging together two SCCS history files this way would still be really fast. To give an impression: entering the current version of OpenSolaris into SCCS (52271 files with 550MBytes in total) takes 15 seconds using a 12 year old Opteron machine; this is 37 MBytes per second... SCCS is the fastest known revision control system. Schily (talk) 11:29, 27 February 2015 (UTC)

As usual - no reliable source mentioned in your comments. Regarding redistributing any code from McVoy, you certainly recall how vindictive he is on that topic. I'll sometime get around to writing about BitSCCS in the context of the tools I wrote a while back, which doesn't require that type of interaction TEDickey (talk) 00:56, 28 February 2015 (UTC)

As usual absolutely nonconstructive blabla. Try to understand what's a reliable source is. I decided not to waste any time with you anymore until you start doing something constructive. BTW: I never introduce something I cannot verify with reliable sources. Reliable sources are however not always shareable with the world. Everybody who invites me for a beer can get a view on the non public sources... Schily (talk) 11:18, 28 February 2015 (UTC)

Well then, why don't you finish your introduction by providing the verifiable, reliable sources? (You can keep your beer). TEDickey (talk) 12:19, 28 February 2015 (UTC)

I do and once you start being fact based and give reliable sources, we can continue. Schily (talk) 17:28, 28 February 2015 (UTC)

By the way, the discussion to this point does not establish that the patent used as a source is synonymous with "smoosh". In particular, the patent disclosure does not mention this. The "better source" tag asks for a source which establishes that connection. TEDickey (talk) 01:50, 12 November 2015 (UTC)

Will you ever be constructive?

Will you ever even describe what your intention is? Note that your previous edits, as usual definitely are no more than just edit wars as you did not even explain why you added an unneeded maintenance tag.

Now that you finally mention what you like, it seems again that you edited an article without having the needed skills for the topic or are not interested in the article but rather in edit warring. Glenn pointed to the patent number for his invention and even the patent about the "wheel barrow" does not mention the term "wheel barrow". If you have some basic understanding of the underlying principles, you are still able to see that it describes a "wheel barrow". The same applies to US patent #5481722. Given that you previously confirmed that you own a copy of an old sourcecode from KitKeeper, it is even easy to compare the smoosh implementation from Larry McVoy with the invention from Glenn Skinner and it is possible to identify the algorithm implemented by Larry McVoy together with a lot of background information that is missing in the uncommented code from Larry McVoy. You just need to read and to understand. Schily (talk) 10:29, 12 November 2015 (UTC)

What you are saying is that you have no published source, just random personal observations. That is not useful for this topic. TEDickey (talk) 02:12, 13 November 2015 (UTC)

The US Patent is no published source? You seem to have a strange view Schily (talk) 10:15, 13 November 2015 (UTC)

As usual, you have no source to discuss. TEDickey (talk) 01:29, 14 November 2015 (UTC)

As usual, you have no sources for your claims Schily (talk) 10:42, 16 November 2015 (UTC)