Talk:Perl/Archive 4

From Wikipedia, the free encyclopedia

 This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
← Archive 3 Archive 4 Archive 5 →

Contents

TCPI?

-Barry- just added a paragraph linking to the TCPI Long Term Trends chart and uses it to claim that Perl is rapidly losing in popularity. This index is nothing more than using Google and other search engines to search for:

+"<language> programming" -tv -channel

This has about as much credibility as the old Operating System Sucks-Rules-O-Meter and I'm just itching to remove or at least rewrite that section. What do other people think? Imroy 06:23, 10 May 2006 (UTC)

I had this same discussion on Perlmonks and in Talk:Comparison of programming languages. Here's what I wrote in the latter:
--------
There's Wikipedia:Search engine test. Under idiosyncratic usage it says "A series of searches for different forms of a name reveals some approximation of their relative popularity."
The TIOBE data has been around for a long time and is regularly updated. At [1], where it says "The definition of the TPC index can be found here" you can click "here" to see an explanation. I don't think it's a good explanation, especially for laymen, but it provides some details that might matter to some people.
Here's a discussion that might be of interest.
--------
If necessary, I can do what I did in Comparison of programming languages and clarify that the data is gathered from search engine results. The numbers are actually more accurate than that sounds, so I may explain it a bit more than that, but I'd try leaving out the long footnote that I put in Comparison of programming languages. -Barry- 07:29, 10 May 2006 (UTC)
The problem is that you have only one data point: the number of times Perl is mentioned on the Web. You don't have:
  • Number of times it's mentioned in journal articles
  • Number of times it's used in deployed systems
  • Number of times it's mentioned on the Web (but not refered to as "programming").
The other problem that I have with the idea of judging Perl "popularity" by Web hits is the idea of maturity. No one is running around saying, "hey look at this cool new language I just found," about a 20 year old programming language. On the other hand, 20 years worth of programmers continue to use it day in / day out as one of the primary tools in their toolbox.
I'm also still a bit concerned about anyone who makes this edit summary comment: [2] and then procedes to push for questionable language comparisons... Please review WP:POINT as soon as possible, and I will continue to attempt to assume good faith. -Harmil 18:28, 10 May 2006 (UTC)
Thanks for pointing me to that badly written guideline. I fixed it.
From various comments I've heard about TIOBE's data and my own data from a similar study, it seems there are people who think it's meaningful, even once they know where it comes from. It's just part of the information in that paragraph that makes the argument that Perl is losing popularity. I've read Perl has lost popularity elsewhere as well, and I'm sure I'm not the only one. If you must edit it, say something like "...based on advanced search engine queries." Maybe add a footnote with the exact query and list the search engines. Unfortunately, I don't think the page on TIOBE's website with the details is directly accessible with a link when Javascript is on. -Barry- 19:32, 10 May 2006 (UTC)
Of course the data has meaning. Almost any statistic has some sort of 'meaning'. You just have to be careful to explain what it's measuring and what it's showing. And I don't feel you are doing that properly. Since it's using simple (NOT advanced) Google/Yahoo/MSN searches, it's probably picking up mostly forums, blogs, and corporate hype. It's not picking up on the number of actual projects using a particular language, or the number of programmers with experience in that language. A search involving job advertisements would be much more realistic, IMHO. This TIOBE index is extremely superficial. Imroy 23:57, 10 May 2006 (UTC)
Not only that, but these statistics do not include any knowledge of what languages are used for intranets which can form a large part of commercial development. Enigmatical 00:07, 11 May 2006 (UTC)

TIOBE gives relative rankings. Perl needn't lose popularity in order to fall in the rankings. All that is necessary is that other langauges grow faster. I have one data point that bears on this. I host the Perl Module Mechanics page. Traffic to this page has been growing slowly, but steadily, over the last few years. Swmcd 23:24, 10 May 2006 (UTC)

The word popularity still fits. I don't think saying something like "per capita" or "relative to..." is necessary. Citing the increased number of users (or web searchers) would be useless unless you cite it for of other languages too. But clarify whatever you want if you think it's necessary. -Barry- 23:36, 10 May 2006 (UTC)

Good grief! Has anyone looked at the TIOBE page? Between 2004 March and 2004 April, Java dropped from 24% to 19% in the rankings. At the bottom, they have this FAQ

# Q: What happened to Java in April 2004? Did you change your methodology?

A: No, we did not change our methodology at that time. 
Google changed its methodology. 
They performed a general sweep action to get rid of all kinds of web sites that had been pushed up. 
As a consequence, there was a huge drop for languages such as Java and C++. 
In order to minimize such fluctuations in the future, 
we added two more search engines (MSN and Yahoo) a few months after this incident.

With artifacts like that, how can we cite TOIBE as evidence of anything? Swmcd 23:45, 10 May 2006 (UTC)

The TIOBE paragraph also says

OSCON — the open source convention sponsored by book publisher O'Reilly — is 
much less Perl-oriented than it used to be. Randal L. Schwartz, author of several 
Perl books published by O'Reilly, has said that OSCON's organizers are openly 
hostile to Perl, and that Perl isn't interesting to O'Reilly anymore.

Could we get a direct quote here? Or a report from Randal himself?

A further question is whether this is relevant to Perl, or only relevant to people who publish books about Perl. Swmcd 23:54, 10 May 2006 (UTC)

I gave the source of the quote in the edit summary. It's from a chat logged at [3]. It comes from more than one sentence, and I figured it would look better not to make it a quote. -Barry- 02:48, 11 May 2006 (UTC)
I thought there was a policy against reposting logs? Also, this edit isn't very informative regarding Perl or opinions on it. It seems more like the edit was made to embarass Randal, O'Reilly, or both. Steve_p 19:49, 18 May 2006 (UTC)
It's related to Perl's popularity and changing opinion of Perl. I don't know about the reposting logs rule, but it's not like it's a personal log. The log is well known to that community and on a public website. Major Perl 6 development takes place on Freenode's #perl6 channel, by the major players, and it's run by the Perl Foundation (or whatever they're called). -Barry- 21:10, 18 May 2006 (UTC)

I think this material needs to be moved to another section. As discussed above under Talk:Perl#Pro vs. Con, the purpose of the Perl#Opinion section is to present and summarize opinions that are held of Perl. Data indicating the prevalance of various opinions isn't really to the point. Compare the American_politics page, which has separate sections titled American_politics#Political_culture (outlining the political views of Americans) and American_politics#Political parties and elections (tabulating recent election results). Swmcd 03:39, 13 May 2006 (UTC)

This is another stupid idea from Barry. The data DOES NOT SHOW that Perl is losing in popularity. Period. It cannot show that, it does not show that, and it is incorrect to state or otherwise imply that it shows that. Pudge 15:22, 18 May 2006 (UTC)

Benchmark comparison

There's already a Comparison of programming languages article, but it wouldn't be practical to include a detailed comparison of every two languages. A detailed comparison between Perl and other languages might work for Perl's article, or maybe for a separate article titled Perl language comparison. I did that "be bold" thing and added it to this article.

I remembered this Perlmonks thread about a contest that I was thinking of mentioning here or in Comparison of programming languages, and that lead to me finding this page of links to various benchmarks. I clicked "Debian : AMD™ Sempron" then clicked "Perl" then selected the languages to compare Perl to, and that's what my tables are based on.

There are disclaimers on that website about the data not always applying, and there are other benchmarks for other operating systems. Maybe more information should be gathered for a separate page. -Barry- 08:24, 11 May 2006 (UTC)

This benchmark is horribly, horribly flawed. In just 10 minutes of cursory analysis, I was able to modify the first benchmark[4] so that it ran over 20% faster in Perl. I've removed the results, as it seems that no useful information can be extracted from such a flawed study. -Harmil 14:39, 11 May 2006 (UTC)
That kind of thing works both ways and probably balances out. Could you post your modification somewhere? -Barry- 18:20, 11 May 2006 (UTC)
I gave you the link to the modification above, and no, we cannot assert that it "probably balances out". Can we please, just stick to facts. This benchmark, like almost any cross-language benchmark is horribly flawed, and should not be referenced. -Harmil 19:21, 12 May 2006 (UTC)
Sorry, I didn't know that was the link to the diff. Unfortunately, I'm not sure how to read that and I don't want to learn. I like the added context of surrounding code anyway. I'd prefer if you use my DiffNote web application and link to the results. But it probably won't make a whole lot of difference in my opinion of whether to add the Benchmark Comparisons section. In fact, I wouldn't even be able to cite your improved version of that benchmark because of the no original research rule. On the other hand, I wouldn't object to a stronger disclaimer, especially if you've actually managed to legally improve the speed, and you may be able to submit your benchmark, which would allow it to be used in this article.
Also, even if you did speed the script as fast as you claim without breaking it, that doesn't mean Perl would have won any more tests, and if it did win one more, that's only a small fraction of the 16 or so speed tests (32 now that I doubled the data). Using one improved test to discount the validity of the entire data set is itself flawed.
I noticed that part of your modification goes against best programming practices. You replaced descriptive variables like $item with Perl's default variables like $_[0]. I guess that's technically allowed, but I wouldn't want to maintain your code.
I think we need more input on this, because I really think the Benchmark Comparisons section improves this article and I'll probably put it back. I don't think there's anything like it in other programming language articles and it could help make this article a featured page.
Again, I wouldn't mind disclaimers and references that show that the results should be considered with caution. In fact, the section can concentrate on the flaws in cross-language benchmarks, but some people would figure it balances out and may put some value on the data. How DO you choose a programming language if you want speed, or something else that can be benchmarked, but don't have different versions of the exact program you'll be using that are written in multiple languages? If you do have some important snippets written in multiple languages, how do you know you wrote them each optimally so the comparison lives up to Harmil's standards? People need to make those language choices sometimes, and most probably base it on less accurate information than is contained in these benchmarks.
It's probably more accurate to use the references here to learn how to create a benchmark specific to your machine and program (I included that link in the Benchmark Comparisons section), but if you don't want to go through all that, what do you do? Ask some guy on a message board? Questions like "what language should I use for..." are asked all the time, and when it comes to needing speed or fewer lines or less memory, the data Benchmark Comparisons article is probably better. -Barry- 20:50, 12 May 2006 (UTC)
Here's a link to the last version of the Benchmark Comparisons section that was reverted, and here's a link to Wikipedia's dispute resolution guidelines, in case it's necessary. -Barry- 21:08, 12 May 2006 (UTC)

The Benchmarks section now carries a Factual Accuracy template, but that isn't really the issue. Assuming that the benchmark results were correctly copied from their source, they are accurate. The question is whether they are meaningful or relevant to people reading the Perl page. This is an encyclopedia article. It is meant to provide a introduction/overview for non-specalists. A link to a site with benchmarks results might be appropriate; an entire section of benchmark results copied from somewhere else probably isn't. And adding a link to a page titled Flawed Benchmarks doesn't redeem it; it makes the case that the section shouldn't be there in the first place. Swmcd 02:48, 13 May 2006 (UTC)

There's a conceptual problem with a Perl vs. C benchmark: the Perl interpreter is a C program. On its face, the Perl vs. C box that gives Perl 1/1 wins for speed asserts that Perl is faster than C on that benchmark. But underneath, what it means is that the Perl interpreter has faster C code than the C implemetation of the benchmark.

We could improve the C implementation by extracting the relevant code from the Perl interpreter and using that instead. But that would probably be difficult and tedious. Here's a better idea: for the C implementaion, take the Perl code, and let the Perl interpreter run it. The Perl interpreter will execute the relevant C code, and now the C implemetation is as fast as the Perl implementation :) Swmcd 03:05, 13 May 2006 (UTC)

The table layout needs work. When I looked at it, I assumed that the numbers in each cell were tests won/total tests. Then I saw entries like 9/8, and knew that couldn't be right. After looking at the table heading, I realized that each cell reports two different counts, in the format AMD/Intel. This needs to be made obvious to the casual reader. Swmcd 03:16, 13 May 2006 (UTC)

Would a comma or pipe be better than a slash? And maybe make a number bold if it's higher than the corresponding number in the other column. -Barry- 04:25, 13 May 2006 (UTC)
-Barry-, please review my recent edits. Note that I have condensed large tables of infromation into fairly simple prose of three paragraphs' length. There are several problems with dumping huge tables AND massive amounts of disclaimer into this article. They clutter the page, aren't really Perl-specific, are arguably not very encyclopedic, and don't actually further the point of the article: explaining the nature of the Perl programming language. I hope that you can respect the need to have a clear and relaitvely usable article for any language, not just Perl, and will continue to edit in that light. -Harmil 22:31, 13 May 2006 (UTC)

Bias being added

Please -Barry-, stop editing this article to insert bias. I've placed an entry for this article on the RfC list. You've been quite clear in your convictions in your edit summaries [5], your talk page [6], and your refusal to remove material which has not only been disputed, but clearly demonstrated to be incorrect. [7] It is time to let it go. -Harmil 12:21, 13 May 2006 (UTC)

A description might help. Your RfC entry is here, under Telecommunications and digital technology, but it just says "Talk:Perl 12:15, 13 May 2006 (UTC)" -Barry- 17:14, 13 May 2006 (UTC)
You didn't sign it either. Read the Request for comment template on top of that page, particularly item 1 and 3. It says:

1. List newer entries on top, stating briefly and neutrally what the debate is about.

2. Provide a link to the relevant article's talk page.

3. Sign entries with the date only, by using five tildes: ~~~~~.

-Barry- 17:22, 13 May 2006 (UTC)

The instructions for the RFC process is horrible. In one location it asks for the talk page link and date only (and gives an edit link), then on the page that you actually edit there are the guidelines you listed which don't match at all, but you never see those, unless you scroll up after the section edit. Doh! -Harmil 21:39, 13 May 2006 (UTC)
So the question remains, -Barry-: do you intend to keep modifying this article to suit your "anti-perl" (your words) sentiments, or will you yield to the request of three (so far) editors who have cited strong concerns with you edits? I'm going to do a bit of editing to demonstrate what I'm talking about. Of course, as a fellow editor, I invite you to revise my edits in ways that help Wikipedia to cover the language, but not to re-introduce bias. This is an article about a programming language, not your preferences. -Harmil 21:54, 13 May 2006 (UTC)
There may or may not be an anti-Perl flavor to my edits, since I'm somewhat anti-Perl but try to be fair at the same time, but I still think my edits are accurate and improve the article. You initiated a RfC, so lets let that happen before removing large amounts of content from the section in question. We'll have a vote after that if you want. This isn't my article and I won't put my version back once enough people have been heard, preferably who aren't currently involved with Perl. -Barry- 22:53, 13 May 2006 (UTC)

I have given Barry warning: no more bullshit. If he continues, I will just routinely revert every edit he makes in Perl, regardless of what it is. Enough is enough. Other editors who are not Barry, feel free to join in the fun. Pudge 15:25, 18 May 2006 (UTC)

Bias being added II

In this edit summary, Scarpia refers to a slide in this document that says "...and it sux..." His edit summary says "it's unclear what the slide in question means since it's a two word summary of something said in the talk. This doesn't pass the verifiability test, and is irrelevant anyway." I believe it's perfectly clear what the slide refers to. The slide before the "sux" slide says "Benchmark.pm comes with Perl..." and the slide after the "sux" slide continues the discussion of the module. It's perfectly clear that "...and it sux..." refers to Benchmark.pm.

Scarpia's edit summary says it's irrelevant anyway. When mentioning Benchmark.pm in a Benchmark section and mentioning the slides, how could it be irrelevant to say a slide says it's a bad module?

Scarpia edited the Benchmark section so it no longer mentions the slide that makes a negative reference to Benchmark.pm. He's only allowing mention of the slide that warns about benchmarks in general. Also note that he recently blatantly vandalized this article by reinserting previous vandalism (the paragraph on the lower right of that diff). Yet another case of clear bias.

I tried to compromise by not using the quote that contains "sux" and instead using "...[the slides] warn people about the limitations of benchmarks — particularly about the module Benchmark.pm". That should be reinserted into the article.

I believe Scarpia is Brian D Foy, so if he openly admits that and if he now believes something opposing what the slides say, or if he wants to clarify something or claim he didn't mean the slide text literally, he should speak up and his new opinion or clarification can be included in the article after the description of the slides. Though his saying he's Brian D Foy may not be enough for some of the editors here. Either way, the slide with the negative reference to Benchmark.pm should be mentioned. -Barry- 02:54, 21 May 2006 (UTC)

"Scarpia's edit summary says it's irrelevant anyway. When mentioning Benchmark.pm in a Benchmark section and mentioning the slides, how could it be irrelevant to say a slide says it's a bad module?"

The reason you can't is because a "slide" does NOT convey what the author meant unless you heard the actual talk the author gave using that slide. In every other instance and to everyone else who wasn't there it is out of context. I happen to know (because the author of that slide has blogged about it) that he wasn't refereing to the module itself but how people misinterpret the results and it was a throw away joke to boot. So your using it is clearly wrong. -sigzero-

Future of Perl 5

It's best for section to have at least two subsections. The Future section should have a Perl 5 and Perl 6 subsection. Perl 6 is already written. Perl 5 should be about the Perl 6 features that are expected to be ported to Perl 5, and stuff. -Barry- 07:28, 13 May 2006 (UTC)

brian d foy's style guide

Brian D Foy apparently wants attention. See [8]

I don't think his name should be in lower case. If so, there needs to be a footnote or something parenthesized to mention his preference and how his name is generally written, and that's sloppy. I'll agree not to use a period after the "D" because maybe it really doesn't stand for anything, but unless there's a consensus for lower casing it, I won't. -Barry- 05:11, 14 May 2006 (UTC)

How a person wishes to be referred to is entirely their right to decide and is not within your remit to alter. If you insist on maintaining this presumptuous and insulting attitude, then you are also authorising us to change your name. I think from now on you should be known as"♥♥♥Jennifer ^_^" instead. -- Earle Martin (talk, contribs) 23:50, 22 May 2006 (UTC)

Al Tobey, President of the Grand Rapids Perl Mongers agrees with me: [9]. -Barry- 21:54, 14 May 2006 (UTC)

Barry wrote in a recent edit summary:
Scarpia, this is a talk page for the Perl article, and I had a
question about notable Perl guy brian d foy, who I now think is you
and who's named in the article. I need to know how to case the name.
Trying to "out" one of the editors now Barry? You're certainly not helping your image in this part of Wikipedia. Imroy 22:28, 14 May 2006 (UTC)
There's good reason to "out" an editor, if I understand what you mean by "out." See Wikipedia:Wikipedians_with_articles which says "Another reason for this page is to notify the community that these Wikipedians are potential autobiographers, with the risks that entails for NPOV in articles relating to them and their work."-Barry- 22:48, 14 May 2006 (UTC)

Call for "See also" links for Pro section

As I've said, I try to be fair. I also like uniformity in the divisions of the article, even if it takes pro-perl material to do it. The Con section has a See also section, and I intend to add another link to it ( http://www.cabochon.com/~stevey/blog-rants/blog-ancient-perl.html ) once it's working again (see http://www.google.com/search?hl=en&lr=&safe=off&rls=GGLG%2CGGLG%3A2005-21%2CGGLG%3Aen&q=%22Algol+made+its+mark+on+the+world%22 ). I found it when searching for pro-Perl opinion. Maybe someone else will have better luck. I'll keep searching for pro-Perl stuff, but I'll probably add the anti-Perl link whether I find anything or not. -Barry- 00:36, 15 May 2006 (UTC)

Please review Talk:Perl#A warning on adding content before hacking on the Perl#Opinion section.
More generally, consider that the purpose of a Wikipedia article is to inform its readers, and that the structure of the article is subordinate to that purpose. Content should be added or rearranged only if it makes the article more useful or accessible to its readers. It should not be done just for the sake of uniformity. Swmcd 02:05, 15 May 2006 (UTC)
Uniformity is nice too though, and I don't want my edits to appear biased because they might be reverted even if they're appropriate.
I was thinking that you might have had the first blockquote in the Con section in mind when you wrote "There is a natural tendency for people reading this section to add material responding to opinions with which they disagree. However, doing so is not appropriate for this section. It obscures the point of the section and risks becoming POV." I almost replaced that blockquote with something more "conny" but I think it's ok. -Barry- 03:03, 15 May 2006 (UTC)

Linking more than first instance if a term

There's some guideline about only linking the first instance of a term for which there's an article, but I rarely find that's a good idea, especially in a technical article in which people are trying to learn something complex. They shouldn't have to search the article for the first, linked, instance, assuming they even know it would be linked, and they shouldn't have to type it into the Wikipedia search box.

It can also look sloppy and inconsistent, like someone forgot to link the term. For example, at Control structures, C isn't linked, but right next to it Java is linked.

Maybe the talk page should have guidelines on top, explaining why every instance of any term that's linked should be linked, so the Wikicops will understand the reasoning--that usability is especially important for articles like this. -Barry- 07:25, 17 May 2006 (UTC)

I don't know if it's mentioned in the guides, but it's not usually a good idea to only link the first instance in a long article. It really should be "once per major section" or something similar. Imroy 07:47, 17 May 2006 (UTC)

References section

I just noticed the Wikibook template in the references section, and now I'm wondering whether the references section is just for references used in writing this article or not. If so, I think the heading should something more clear than "References." If not some of the links in that section should be under External links. -Barry- 08:18, 17 May 2006 (UTC)

Popularity, again

I removed the following:

There's also criticism of a less technical nature that may be no less important to some. Some people believe Perl's popularity has declined. in the years betwen 2001 and 2006. This may represent a true decline in Perl's usage or an aspect of the transition from a young to mature language. It may also be a result of the fact that it has been well over a decade since the last major release of Perl, and Perl 6 development continues after over 5 years without an initial relase.
In addition, OSCON — the Open Source Convention sponsored by O'Reilly, which started as The Perl Conference — is much less Perl-oriented than it used to be. Randal L. Schwartz, co-author of several Perl books, has said that OSCON's organizers are openly hostile to Perl, and that Perl isn't interesting to O'Reilly anymore.(Paraphrased from merlyn's posts (Randal Schwartz's IRC handle) on Freenode's #perl6 irc channel. sel=86#l151 Archive of chat; sel=93#l161 continued archive)

The first paragraph is classic weaselling without any facts to back it up. The second is irrelevant - what OSCON chooses to focus on is not the measure of Perl. Personal opinions expressed about what seems to be the opinion of someone else (in this case, the OSCON organizers) are not encyclopedia material. -- Earle Martin (talk, contribs) 14:18, 19 May 2006 (UTC)

Facts to back up opinions aren't necessary in an opinion section, but it's true that there needs to be some indication of what people's opinions are. I don't think that the quality of documented evidence needs to be very high if most of us have heard, as I have, that Perl isn't as popular as it used to be. Whether you've heard that or not, there was some documented evidence indicating that some people would believe Perl is losing popularity, but it was reverted. It was the part about the TCPI Long Term Trends chart in the following paragraph.
"There's also criticism of a less technical nature that may be no less important to some. Perl's popularity has declined. As of May, 2006, the TCPI Long Term Trends chart of the ten most popular programming languages shows that Perl's popularity is at its lowest since before June, 2001 (the earliest date plotted), and has dropped more than any other language over the past year. In addition, OSCON — the open source convention sponsored by book publisher O'Reilly — is much less Perl-oriented than it used to be. Randal L. Schwartz, co-author of several Perl books published by O'Reilly, has said that OSCON's organizers are openly hostile to Perl, and that Perl isn't interesting to O'Reilly anymore."
Here's an excerpt from Wikipedia's weasel words page:
"Weasel words are words or phrases that smuggle bias into seemingly supported statements by attributing opinions to anonymous sources. Weasel words give the force of authority to a statement without letting the reader decide if the source of the opinion is reliable."
With the TCPI reference, readers could decide for themselves whether the belief that Perl is losing popularity has any basis. Without it, assuming it isn't common knowledge that Perl is losing popularity, you may have a point. I don't like the trend in recent edits to this article to remove information that there's some controversy over rather than explaining things, especially in the opinion section (the related links, for example), and I'd prefer to see the TCPI reference brought back.
About OSCON, I think O'Reilly's opinion counts, and the opinion of Randal Schwartz counts. -Barry- 02:28, 20 May 2006 (UTC)
I think a good rewording of the first paragraph would simply be:
"There's also criticism of a less technical nature that may be no less important to some. Some people believe that Perl's popularity has declined. As of May, 2006, the TCPI Long Term Trends chart of the ten most popular programming languages shows that Perl's popularity is at its lowest since before June, 2001 (the earliest date plotted), and has dropped more than any other language over the past year, though the relevance of the chart, which is based on search engine results, is disputed. In addition, OSCON — the open source convention sponsored by book publisher O'Reilly — is much less Perl-oriented than it used to be. Randal L. Schwartz, co-author of several Perl books published by O'Reilly, has said that OSCON's organizers are openly hostile to Perl, and that Perl isn't interesting to O'Reilly anymore."
I added "Some people believe" and "though the relevance of the chart, which is based on search engine results, is disputed." Didn't add it to the article yet. -Barry- 02:42, 20 May 2006 (UTC)
Barry, just stop. You clearly have an anti-Perl agenda, you even admitted so yourself. That TCPI chart has been shown to be unreliable at best, and any attempt to dress it up is just an excercise in weasel words. You've worn out your welcome here. Go away and stop adding your thinly-veiled bias to this article. Imroy 03:01, 20 May 2006 (UTC)
I'll say again what I've said three times, regarding the relevance of search engine results like the ones used in TIOBE's data:
I had this same discussion on Perlmonks and in Talk:Comparison of programming languages. Here's what I wrote in the latter:
--------
There's Wikipedia:Search engine test. Under idiosyncratic usage it says "A series of searches for different forms of a name reveals some approximation of their relative popularity."
The TIOBE data has been around for a long time and is regularly updated. At [10], where it says "The definition of the TPC index can be found here" you can click "here" to see an explanation. I don't think it's a good explanation, especially for laymen, but it provides some details that might matter to some people.
Here's a discussion that might be of interest.
--------
I'm not dressing up anything. In fact, I've dressed the TIOBE chart down for you, and note that I haven't added it to the article yet. Likewise, please don't delete large amounts of the article without prior discussion. The existence of the chart, its backlinks, and its Google PageRank of 7 make it relevant, and I won't even object to adding your own point of view about it to the article, because it looks like there's no getting around that.
-Barry- 03:28, 20 May 2006 (UTC)
Barry, the Wikipedia:Search engine test is used for a number of things, mainly to decide if an article is noteworthy or should be deleted. Nowhere on that page does it say anything about using the relative result counts from seperate searches for different things. Nowhere does it say anything about using those statistics to conclude that something is losing popularity. You are hiding behind that page as if it defends the information you wish to introduce or the conclusions you have drawn. It does not, and Craig Stuntz even said so on the Talk:Comparison of programming languages page. Can you please just drop the TIOBE chart/data? I'm not going to accept its inclusion in the article as anything but a minor curiosity. Imroy 05:08, 20 May 2006 (UTC)
-Barry- wrote
Facts to back up opinions aren't necessary in an opinion section
This is false. There's nothing special about the Perl#Opinion section. It undertakes to report some facts about Perl, vis. opinions that people hold of it. Those facts must be supported by evidence and references, just like the facts reported in any other section.
I don't think that the quality of documented evidence needs to be very high
if most of us have heard, as I have, that Perl isn't as popular as it used to be.
Also false. The quality of evidence required for an assertion in a Wikipedia article does not depend on how many editors have heard the assertion. Swmcd 04:20, 20 May 2006 (UTC)

Imroy: I think you're being too literal, but I think just the fact that the chart exists and the reason for it that's mentioned the website, and its popularity shows that changes in search engine results are a consideration to some people. I disagree with Craig Stuntz too. I think he's the one who reverted the part of the footnote in Comparison of programming languages that said the TIOBE data an approximation of popularity, even though the full details about the origin of the data was right there in the footnote to avoid misleading anyone. That was a bad reversion, but I let it stand. At least the data's there.

Swmcd: point one, then it's not about opinion and it's just Pro and Con, and the heading Opinion shouldn't be used. I did let Randal's reversion of one of the links stand without questioning him about the problem because I believe he'd know whether something in a Perl article is inaccurate and I wouldn't want a high level of clear inaccuracy, but that's probably biased of me and the link should probably be included considering the limited number of such links that are available.

Swmcd: point two, I don't think there are enough sources given on Wikipedia, but I disagree that whether something is common knowledge to the editors has no bearing on the amount of evidence required. -Barry- 06:08, 20 May 2006 (UTC)

Bottom line on the popularity thing: popularity (or lack thereof) does not constitute a "criticism" of the lagnauge. Not in any way. I'm also not clear on how popularity even fits in as a notable item for a Wikipedia entry, other than to say that Perl is one of the more popular programming languages out there. Speculation beyond that would require visibility into many industries and many development processes. It's also impossible to separate enthusiasm from usage, which further makes popularity an unquantifiable item. Mature languages tend to have much less enthusiasm, this much is clear, but I'm not sure that there can be a reasonable argument that Perl (or any other language's) usage has moved up or down, as most companies don't disclose. -Harmil 22:13, 21 May 2006 (UTC)