Talk:Ian Jackson

Latest comment: 11 years ago by Σ in topic dpkg Authorship

Untitled

edit

I am no longer the chairman of the Debian TC. That post is now held by Bdale Garbee. I'm still a member of the TC. -iwj — Preceding unsigned comment added by 212.13.197.229 (talk) 10:00, 18 September 2009 (UTC)Reply

dpkg Authorship

edit

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.


Statement of the RfC problem?

edit

I came here in response to an RfC thinking that I would find a sentence or two explaining why the RfC was sent out. I can't find any. Let's put a concise explanation at the beginning of this, right below the RfC label. DonaldRichardSands (talk) 11:14, 26 October 2012 (UTC)Reply

See “Summary” below. It seems have become detached from the label in subsequent edits. Ewx (talk) 08:06, 27 October 2012 (UTC)Reply

Authorship history, redux

edit

The C version was not a simple translation from Perl to C. The C version contains a great deal of functionality which was missing or broken in the Perl script. The perl script was only ever intended to be a stopgap. In particular, the Perl script just used "tar" to unpack the filesystem archive whereas the C program used an existing miniature tarfile reader library (from Bruce Perens IIRC) to handle each file individually in the correct manner. Another aspect is that the majority of the code in dpkg is error handling of one kind or another; the Perl script didn't really take much care over the consequences of errors.

There were IIRC two script versions of dpkg; the first one I inherited from Ian Murdock but may well have had other contributors. If I remember I rewrote the first script to make a better stopgap. The second rewrite was the one which involved rewriting the whole thing in C. It involved significant new design decisions including the current sequencing of filesystem actions and package scripts, and it involved documenting those design decisions in what was then called the dpkg Programmers' Manual.

The source package dpkg contains both the program dpkg, and the utilities for manipulating source packages. Those utilities I did not entirely rewrite, although during my maintainership I did make substantial changes and introduced new utilities.

The above is true to the best of my knowledge but it's been many years. If someone wants to verify the facts independently the right way would probably be to dig out relevant ancient versions of the source and to compare the Perl and C versions.

Ijackson (talk) 13:09, 1 October 2012 (UTC)Reply

I see the page currently says something about me having "ported" dpkg from Perl to C. This is not an accurate statement. The program needed to be rewritten and at the same time it was convenient to change the implementation language.

Wikipedia's page Porting explains that porting has the purpose of allowing a program to be run in a different environment; I agree with that use of the term. The environment of the Perl and C versions of dpkg - the Debian GNU/Linux system - was the same in both cases.

As I have written above, the C version was a rewrite. It was certainly not a line-by-line or even function-by-function translation. The statement that it was "based on" the Perl version (as you see in the dpkg AUTHORS file) is fair although of course very vague. It was "based on" the Perl version in the sense that it used most of the same data formats and performed roughly the same functions although with significant differences eg in ordering of operations.

Ijackson (talk) 14:20, 2 October 2012 (UTC)Reply

Summary

edit

The discussion is about the authorship of dpkg. dpkg as it stands is a C program with well over 100 contributors.

The anonymous user at 86.176.68.157 claims that the transition from its Perl predecessor to a C program (which was done by the subject of this article) was merely "porting" from one language to another, analogous to translating a novel. I’m not aware of any reference having been provided for this view.

The other view is that the C version was a substantially new program with a much greater feature set. This view is supported by dpkg's own AUTHORS file, which lists the subject of this article as its original author, and describes it as "based on" the Perl version - not merely translated or ported from it.

Ewx (talk) 08:13, 26 September 2012 (UTC)Reply

I have removed a claim by 86.* that IWJ lied in the AUTHORS file about the authorship, under WP:BLP: it is potentially libellous. Marnanel (talk) 13:44, 1 October 2012 (UTC)Reply
No claim that he lied, merely that what was put was not factually correct. I have reverted back to the facts, and added the additional information that as well as the port, he added significant new functions, as he himself admits above. — Preceding unsigned comment added by 86.178.183.167 (talk) 12:45, 2 October 2012 (UTC)Reply


Previously uninvolved editor here from FRS. What do the sources say? I am familiar with how porting works, and the sources currently cited in the article seem to agree with the conclusion that he performed a sufficient level of work to be called the author of the C version. Although I'm open to reviewing sources that back up the claim that it was a simple port. Andrew (talk) 23:23, 1 November 2012 (UTC)Reply

Prior Discussion

edit
He also didn't write dpkg - He ported it from Perl to C. The real authors are detailed on the dpkg wiki page.
Translation of a software program from one language to another is no more authorship than translating Harry Potter to Latin would be. — Preceding unsigned comment added by 86.176.68.157 (talk) 16:25, 17 September 2012 (UTC)Reply
For the benefit of anyone confused by the above bizarre claim and the recent vandalism, it’s spillover from a usenet argument which the anonymous contributor has decided to bring to wikipedia. Ewx (talk) 08:09, 18 September 2012 (UTC)Reply
No argument involved. Wikipedia itself tells us that dpkg was written by someone else. — Preceding unsigned comment added by 86.178.183.167 (talk) 18:39, 25 September 2012 (UTC)Reply
Compare for example the implementation of the --list and --status options. The structure of the two implementations is entirely different. I don't really see how anyone could think that some kind of essentially mechanical translation had been involved. Even a superficial look shows that this was a new program. Ewx (talk) 15:02, 2 October 2012 (UTC)Reply

Older Discussion

edit

chiark does not host MindTerm. I just provided a copy of the source as required by the GPL. Also, I have quit nCipher and as of 23.8.2005 I work for Canonical. -iwj

RfC responses

edit

I've added a new section for RfC responses, since there seems to be some out-of-order editing going on here. If you are responding to the RfC, please add your responses below to avoid confusion. Abhayakara (talk) 18:51, 3 October 2012 (UTC)Reply

  • Use the Source, Luke!—there is only one source cited here, and it is a primary source that says that "The C implementation of dpkg was originally written by Ian Jackson." So the text of the article should say something very like this, and no more. Ian's recollections here on the talk page are not WP:RS and cannot be used. Ian's detractors' recollections here on the talk page are likewise not WP:RS and cannot be used. Since the question is disputed, we have to rely on the one source we have (a quick google search turned up no other sources). If someone disputing the current text wants to change it, they must find a reliable source that contradicts the existing source. Abhayakara (talk) 18:51, 3 October 2012 (UTC)Reply
such as the definition of dpkg on wikipedia which states that it was written by someone completely different. Jackson only rewrote this in C from the perl original scripts. This is confirmed very clearly at http://fts.ifac.cnr.it/cgi-bin/dwww/usr/share/doc/dpkg/AUTHORS. As such, rewrote and added to the original is more correct than "authored". — Preceding unsigned comment added by 86.130.68.60 (talk) 20:58, 5 October 2012 (UTC)Reply
      • That's one response to the RfC saying "If someone disputing the current text wants to change it, they must find a reliable source that contradicts the existing source." That you can interpret this as "I am right to change the page" is utterly remarkable. dpkg isn't a source in and of itself. Pinkbeast (talk) 13:13, 6 October 2012 (UTC)Reply
Wikipedia is never a reliable source. Wikipedia relies on reliable sources. If the wikipedia article on dpkg says something that is true, it should refer to some source that says it. You can use that source in this article, but not the wikipedia article that relies on the source. However, as it turns out, the claim of authorship in the dpkg article is unsourced, whereas the claim here is sourced. So the claim here wins. If you think the claim is wrong, come up with a reliable source that contradicts it. Abhayakara (talk) 11:53, 9 October 2012 (UTC)Reply
http://www.debian.org/doc/manuals/project-history/ch-detailed.en.html appears to be a useful source, and hopefully should be official enough to be a Reliable Source, but I am not an expert on wikipedia policy. It contains a quote from Ian Murdock saying "Most of 1994 was spent organizing the Debian Project so that others could more effectively contribute, as well as working on dpkg (Ian Jackson was largely responsible for this)." Also, further down, it contains: "The earliest packaging tool, written by Ian Murdock and called dpkg, created a package in a Debian-specific binary format, and could be used later to unpack and install the files in the package. (paragraph break) Ian Jackson soon took over the development of the packaging tool, renaming the tool itself dpkg-deb and writing a front-end program he named dpkg to facilitate the use of dpkg-deb and provide the Dependencies and Conflicts of today's Debian system." It is clear from this that the authors of the project history consider it fair to say "Ian Jackson wrote dpkg", this is in accordance with the other cited sources. Furthermore, the current statement is deficient on at least two counts. Firstly, it states that Ian based "his work" on "he original PERL code written by Ian Murdock and others." I see no citations for this (the dpkg release notes say "The C implementation of dpkg was originally written by: Ian Jackson <ian@chiark.greenend.org.uk> based on the Perl implementation by:..."), I do see citations showing he made considerable original contributions. Secondly, it should be "he" and not "the", and "Perl" and not "PERL". I am not a regular wikipedia editor, and this looks like a complex matter, so I will leave making the appropriate changes to someone else. 188.39.80.202 (talk) 13:35, 9 October 2012 (UTC)Reply
Good catch, 188.38.80.202. Pinkbeast (talk) 15:36, 9 October 2012 (UTC)Reply

AFAICT, Jackson did substantially more than "simply port" in the case at hand - the sources all seem to concur that his version had substantial changes and improvements. To suggest otherwise is contrary to logic. I would not say "wrote" is the best word - perhaps "substantially improved and extended dpkg in the process of rewriting it in C" would work? Collect (talk) 11:27, 12 October 2012 (UTC)Reply

In my view that is not quite right. From the link above (and my own understanding), the confusion here is that the Perl tool called dpkg largely became the tool now called dpkg-deb, not the C tool called dpkg (which invokes dpkg-deb). So the real issue is that there are two things called dpkg, one written by imurdock, one written by ijackson, but they aren't the same thing. Pinkbeast (talk) 03:16, 15 October 2012 (UTC)Reply

I'm not quite sure how to weigh in regarding the technical discussion, but I'd like to see this article actually turned into a biography rather than a list of technical accomplishments associated with a particular individual. In its current form, I'd be hard pressed to call this article biographical. Unless someone is willing to write a properly cited biography, I suggest that this article be merged into dpkg --Scalhotrod - Just your average banjo playing, drag racing, cowboy... (talk) 15:32, 18 October 2012 (UTC)Reply

Of course I'm biased here, but I think that would be a mistake. It would, for example, create a curious discontinuity in the list of Debian Project Leaders. 81.187.27.126 (talk) 15:37, 18 October 2012 (UTC)Reply
Then NONE of these articles are biographies and their format is poorly thought out and composed. I have nothing against documenting these individuals, but this and the related articles are not biographies as evidenced by your point about the list of Debian Project Leaders.
It appears that a series of articles have been created that should be an enhanced list rather than this long, awkward chain. --Scalhotrod - Just your average banjo playing, drag racing, cowboy... (talk) 15:45, 18 October 2012 (UTC)Reply
I'm not as familiar with the technicalities of this - so factor that in when reading this - but it sounds like Jackson should be considered the author and that he did do major work to deserve this distinction.Dreambeaver(talk) 19:20, 19 November 2012 (UTC)Reply

Edit war

edit

86.* has threatened to keep reverting changes while they're logged on. Are we now in an edit war, and specifically, has WP:3RR been infringed yet? Marnanel (talk) 13:10, 6 October 2012 (UTC)Reply

I have stated that I will continue to revert this vandalism. Jackson did not author dpkg, and the only two sources available (dpkg and Wikipedia) demonstrate this. The fact that Jackson and any of his supporters disagree is not a valid reason for their constantly making the false claim that he wrote it when he himself admits he did not. Surely someone should invite the original PERL authors to comment on whether they feel Jackson "wrote" the package or just took their ideas and reimplemented them in a different language.
So, "yes" then. Marnanel (talk) 20:06, 6 October 2012 (UTC)Reply
By trying to keep the entry accurate, I am not vandalising it. — Preceding unsigned comment added by 86.130.68.60 (talk) 22:34, 6 October 2012 (UTC)Reply
No, this is not how Wikipedia works. "Trying to keep the entry accurate" is not a defence against WP:3RR. In any edit war, everyone thinks they're trying to keep the entry accurate. Marnanel (talk) 00:40, 7 October 2012 (UTC)Reply
Then *please* provide a verifiable third party source of authorship. Even Jackson admits he started from someone else's code-base. — Preceding unsigned comment added by 86.130.68.60 (talk) 11:28, 7 October 2012 (UTC)Reply
Provided references to debian-devel; Bruce Perens stating dpkg is Ian Jackson's design; Ian also clearly referred to as "the author". Pinkbeast (talk) 13:39, 7 October 2012 (UTC)Reply

However, with or without a third-party source of authorship, my point stands: irrespective of who's in the right here, edit wars are not allowed on Wikipedia. Marnanel (talk) 15:11, 7 October 2012 (UTC)Reply

As an observer here, since there is no third-party reference to Dr Jackson writing dpkg, and the sources cited are not verifiable, it is clear that any reference to his authorship, re-writing, porting or translating dpkg ought to be removed entirely until there is such a source. Just my POV. Tramlink 21:07, 7 October 2012 (UTC)
The AUTHORS file was added to dpkg by Scott James Remnant. Isn’t he a third party? Ewx (talk) 08:07, 8 October 2012 (UTC)Reply
I think the current entry accurately describes the situation. PLeased to be of service Tramlink 09:17, 8 October 2012 (UTC) — Preceding unsigned comment added by Tramlink (talkcontribs)
Pretty sure that posting to dpkg-devel that refers to Dr Jackson as "the author" was written by a third party. Also quite sure Bruce Perens is a third party. It's nice of 86.* to log in, but whether User:Tramlink edits anonymously or no, it is not reasonable to take three sources two of which identify Dr Jackson as the author and one which describes dpkg as his design and produce the current mish-mash, which is quite clearly grinding the same axe. Absent further comment I will edit the page to be consistent with the sources tomorrow. Pinkbeast (talk) 17:30, 8 October 2012 (UTC)Reply
It currently says "the first C implementation" which is rather odd because there has only ever been one C implementation and it's the one currently in use. This is a very clumsy way of acknowledging the changes made by subsequent maintainers. I would suggest "He also originally wrote the C implementation of..." or something along those lines, or some similar wording that acknowledges that the version I wrote is still in use but has seen substantial contributions by other people since then. Ijackson (talk) 11:07, 8 October 2012 (UTC)Reply

AS it is not in dispute that you wrote the (base of the current) C code, and you based this on Ian Murdock's Perl scripts, I have modified the entry to cover both of these. Tramlink 11:39, 8 October 2012 (UTC) — Preceding unsigned comment added by Tramlink (talkcontribs)

Personally I think the comment about "based on" is inappropriate in the context of this article. It would be fair enough in the article on dpkg, but in this context it appears to be suggesting more than it says. In particular, it gives the impression that the C version was less than a rewrite. I'm sorry to get pernickety about this, and I know I'm the subject of the article so I have a CoI, but it seems to me that this is an effort to find a "compromise" between an accurate and fair representation of the facts, and the tendentious and malicious user 86.*. In particular, the malicious user 86.* wants to use this WP article to back up their arguments elsewhere that my claims to have written dpkg are "lies". Ijackson (talk) 16:15, 8 October 2012 (UTC)Reply
Are you then saying that you wrote dpkg from the bottom up without sight of the original PERL scripts? And that some of the processes done by the scripts and still done broadly the same way by the C code are completely different? If not, then the position you are taking does, IMHO, seem to be attempting to increase your role in this. Ultimately, as you have conceded, you re-wrote dpkg in C following sight of the original scripts. I cannot see a fairer way of describing this, and, with respect, you are clearly not neutral in this, as the article is about you. Tramlink 16:45, 8 October 2012 (UTC)
  • This entire discussion is bogus. Use reliable sources. Stop arguing about what you personally believe, because it's not relevant. If Jackson really isn't the author of the C version of dpkg, why hasn't someone taken it out of the source code for the C version of dpkg? Why argue about this here? Go to the guys at Debian and have this debate there. Abhayakara (talk) 11:53, 9 October 2012 (UTC)Reply

It should perhaps be pointed out that this is part of an existing Usenet campaign by someone with a coincidentally similar taste in online handles and obsession with the subject of this article to Tramlink. https://groups.google.com/group/uk.net.news.config/msg/5277113ec4536dc9?dmode=source&output=gplain&noredirect&pli=1 https://groups.google.com/group/uk.net.news.management/msg/971b59be8ec91127?dmode=source&output=gplain&noredirect — Preceding unsigned comment added by Ewx (talkcontribs) 08:03, 10 October 2012 (UTC)Reply

Someone who, unsurprisingly, uses the same ISP as 86.*, from the body of the first article linked there. Pinkbeast (talk) 17:45, 10 October 2012 (UTC)Reply
The entry is still inaccurate. did Jackson come up with the idea of dpkg, or did he stand on the shoulder of giants? If the latter, he did not write it. 86.130.68.60 (talk) 22:08, 14 October 2012 (UTC)Reply
He did more than "translate" a complete creature - the additions and changes do, indeed, qualify as "writing." Cheers. Collect (talk) 22:16, 14 October 2012 (UTC)Reply
I eagerly await your efforts to correct all the other thousands of Wikipedia pages that say someone "wrote" a piece of software that was not a completely original idea. Start with Emacs? Pinkbeast (talk) 03:34, 15 October 2012 (UTC)Reply
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Bogus "minor" edits

edit

Please do not mark an edit as "minor" when it is not. The recent edits to this page misuse the "minor edit" facility. — Preceding unsigned comment added by Pinkbeast (talkcontribs) 18:00, 8 October 2012 (UTC)Reply