Beefy Boxes and Bandwidth Generously Provided by pair Networks
"be consistent"
 
PerlMonks  

Section proposal: Best CPAN Modules

by ptum (Priest)
on Jul 27, 2006 at 15:39 UTC ( #564134=monkdiscuss: print w/ replies, xml ) Need Help??

I've noticed a recent trend of folks expressing some frustration with the abundance of choices in terms of CPAN modules, and wishing for a way to separate the wheat from the chaff (or perhaps my metaphor would better be, 'separate the wheat from the quinoa'):

... and I'm sure there are lots more. Let's face it -- most of us are busy people and can not spare a lot of time browsing through CPAN. Heck, I'm feeling guilty about taking the time to write this node. While I am quite gleeful about the fact that CPAN nearly always has a module that helps me do what I need, it is an unavoidable truth that I don't have the time or memory to really stay up-to-date on the 'best' solutions.

Without wanting to start a flame war or hurt anyone's feelings, it seems evident that for some tasks, some modules are 'better' than others, depending on how you define 'better'. Here are a few possible criteria:

  • Simplicity
  • Extent of adoption (popularity)
  • Stability & maturity
  • Portability
  • Performance
  • Minimal dependencies on other non-core modules

Not wanting to duplicate the Phalanx Project, I wonder if we at PerlMonks could come up with a reasonably compact and dynamic list of the 'best' modules available? Perhaps a section of the site could be dedicated to 'most popular modules', where monks could propose an addition to a limited list, and could vote on the usefulness of the modules for a particular task. Or something like Saints in our Book such that a module only makes it onto the list when it achieves a certain minimum popularity. Alternatively, I envision something like the Q&A section, but in each subcategory, there is a list of modules (ranked by popularity) that might apply in that context.

Of course, there is a measure of kwalitee already in place, but it doesn't seem widely known or used (or perhaps I expose my ignorance).

I think there are many monks who might be willing to rank modules with a simple vote, who don't really have the time or willingness to write a module review (although you could argue that if a module hasn't attracted a review, it might not be one of the 'best'). We only have about 150 module reviews, though, and there is no visible ranking (apart from node popularity if you spend a vote yourself to see).

What do you think? What are the 'best' modules from your perspective? I'll start with a list of those that I personally use and like best (in no particular order), fully expecting others to disagree:

Other obvious modules I have missed?


No good deed goes unpunished. -- (attributed to) Oscar Wilde

Editted by demerphq to fix rendering of table on picky browsers

Comment on Section proposal: Best CPAN Modules
Re: Section proposal: Best CPAN Modules
by jgamble (Pilgrim) on Jul 27, 2006 at 16:14 UTC

      Heh. Yeah, you're right, although I do quite a bit of work with Oracle as well. But this underscores the fact that a single person can't really come up with a universal 'best' list of modules -- it is the kind of thing that one needs a community (like this one) to achieve.

        Given that every user has a personal favourite which might not apply to other users it seems logical that over time "all" modules would have to be included in the list.

        Because for any given module $foo there is almost certainly a user or two who believe that is the world's best module!

        For that reason above all I think that such a list wouldn't be as useful as you'd imagine:

        • 20,000 people will enter DBI.
        • 10,001 HTML::Template
        • 10,000 Template::Toolkit
        • and then 100 people each entering a distinct module!

        What I would suggest is twofold:

        1. Add a rating to the module review section and use that in preference to a new section.

        2. When you say you don't have time to keep up with CPAN that is reasonable (and common!) but I think it is almost mandatory to have a quick look when working with a new area.

        Taking you as an example you say you do mostly "databasey" things. I'd not expect you to look at every type of module available every day, but I would imagine that you would have a quick look if you wished to start workgin with say CVS code.

        Does that make sense? Having the "best module" cover some modules for working with CVS servers might stick in your mind, but I'm sure that people who know of CPAN and are working on a new area would be liable to search there first, even if they weren't sure of what was available.

        Steve
        --
Re: Section proposal: Best CPAN Modules
by jdporter (Canon) on Jul 27, 2006 at 16:39 UTC

    As I said over there, any satisfactory solution would need to be dynamic and collaborative; and PerlMonks' existing features don't facilitate this too well. Probably the best we can do here is our wiki node type, of which we have several extant examples, such as this one. I'm not sure how good the locking is on these; it might become an issue if lots of people are trying to contribute.

    Using threads in sections (such as this one) is bound to fail, due both to the non-collaborative nature and the intrinsic obsolescence of ordinary writeup nodes.

    We're building the house of the future together.

      I fully agree ... the primary intent of this thread is to generate interest in a new section. The use of the thread to collect 'popular' modules is simply an effort to (a) show that there are some modules that are universally popular, and (b) get a start on defining the first candidates for the 'best modules list'.

      I'm not sure I understand why you think a dynamic ranked list like Saints in our Book would not work ... ?

Re: Section proposal: Best CPAN Modules
by samtregar (Abbot) on Jul 27, 2006 at 17:08 UTC
    Please, don't hamstring CPAN competition. CPAN is so strong in part because it is so open to new solutions to old problems. Since there's no annointed solution, new modules quickly find users, regardless of the competition from the established players.

    Sure, sometimes this is bad for the individual - inexperienced users sometimes make bad decisions and pick something unstable - but it's good for CPAN as a whole. And with time users of CPAN learn to find the right module for a given task, on a given day.

    Ask yourself this, would DBI be where it is today if oraperl had been listed on the opening page as the "Best way to Program Oracle with Perl"? Would Text::CSV_XS be so bitchin' if every search for "csv" on search.cpan.org terminated in a recommendation for Text::CSV? Of course you're not suggesting either of these, but I think the motivation is similar, as is the likely result.

    I've always thought the CPAN FAQ on a closely related topic had a great response - it simply points to this page:

        http://xxx.lanl.gov/help/faq/statfaq

    -sam

      Wow, I had not really thought out the philosophical implications of a ranking system for modules -- thanks for your insight.

      While it is certainly true that PerlMonks is the center of the intellectual universe, I can't help but wonder if your concern might be overblown. Listing of a module in a PerlMonks 'Best Modules' list seems more likely to generate competition than to suppress it, if I am correct in my assessment of many of our local monks.

      I can imagine CPAN contributors flocking to PerlMonks and shamelessly promoting their modules through RFC posts in search of the fame and fortune that will naturally result from being included on our list. :)


      No good deed goes unpunished. -- (attributed to) Oscar Wilde

      Actually CPAN modules are already ranked - by how often they are mentioned on PM for a start. Popular modules are simply, more popular. That doesn't mean that they are best, for some value of best, just that they are more popular.

      That is good news for someone looking to solve a problem. Popular choices are often easy to find out about and once found are easy to get information about. It's bad news for new modules that may do a better job than the current popular module for the job but haven't yet reached user critical mass - i.e., become popular.

      Having a cohesive section that lists the popular modules and a discussion of each module allows the best of both if the discussion can provide links to up and coming solutions to the same problem the module under discussion solves.

      You never know, with such a section Data::Dump::Streamer may even become more popular than Data::Dumper.


      DWIM is Perl's answer to Gödel
      Don't forget that 3 of the 4 nodes sited by the OP were talking about helping relative beginners. Yes PM is for helping eachother in general, but I think it has a special mission to help those who are beginning to find their way, whether in Perl generally, or in some area that's new for them.

      As grandfather points out in his node further down, most modules are best for something. I agree, it could hamstring innovation to pre-choose a list of "best" modules (going well beyond a set of reviews of module). Let's keep the variety for those who can best make use of it, but provide help to those who would be overwhelmed by it. Each area has one or two modules that go a little easy on beginners and occasional users, usually nat the expense of features, performance or flexibility. A Good Beginnings list would be very useful, and a name like "Good Beginnings", or some such, would tell people that this is just where to start, not where to stop.

      We can have our cake and still make sure that new users have a good supply of solid bread, not just some cake crumbs. (Err... maybe the French Revolution isn't the best metaphore here.)

      Anyway, it's good for the health and prosperity of Perl to make it easier for new and talented people to get things done using it.

Re: Section proposal: Best CPAN Modules
by eric256 (Parson) on Jul 27, 2006 at 17:32 UTC

    An idea to implement this given current monestary facilities (or at least mostly ;) )

    Create a new section "Module Reviews". In this section the module name is used as the title and maybe part of the synopsis or just a link to CPAN is the body. Then we use the keywords nodelet to 'tag' these modules with things related to them. Add a simple directory of that category and use the existing rep system so that those modules can be ranked by rep. (You could always alter the owner as is done in Q&A or something so that it isn't a personal rep battle.) Then also add an index that is a sort of tag cloud that would categorize modules. To me this sounds like a nice thing, as sometimes its hard/impossible to find module on CPAN unless you know exactly what to search for, with a nice tag cloud you could hopefully make finding related modules easier as they would have related tags even if there names varied greatly.

    Would this requrie some modifications? Yes. But the modifications would all add value and are not 100% necessary for such a section to function. Personaly I also think this is different than a modules review section as reviews are a single persons oppions. These Nodes could server as a home for conversations about the modules in question and maybe reviews could be tacked onto the original node, or linked to in replies to it (maybe thats where the wiki idea someone else mentioned comes in, so that everyone can add links to the main body).


    ___________
    Eric Hodges
      ++ To Ericís idea of a Module Review its a great idea. I have thought for some time that it would be very helpful to have the Brethrens personal choices of Module listed and why they have found them useful and in what way they have used them
Re: Section proposal: Best CPAN Modules
by duckyd (Hermit) on Jul 27, 2006 at 21:04 UTC
    While not related to perlmonks, I'm surprised no one has mentioned AnnoCPAN. Also, the ratings on http://cpan.org can be helpful to those trying to determine which module to try first for a given task.
Re: Section proposal: Best CPAN Modules
by GrandFather (Cardinal) on Jul 27, 2006 at 21:30 UTC

    The pertinent question is not so much "What are the best CPAN modules", but "What is the best CPAN module to perform some specific task".

    The answer to the second question is very often, "All CPAN modules are best for some specific task". After all, the authors of modules spend the time to write and publish their modules because the modules provide a facility or way of solving a specific problem that didn't seem to be available when the modules were conceived.

    Sure, there are plenty of unmaintained and even just plain broken modules around, and there are some that are clearly well written and solve their particular problems very well. But a list of the "best" is just barely usefull.

    Of much more potential use is to change the focus slightly and look at "The most popular modules". Those are the ones that are used a lot, often because they are the best, but also because the solve problems that are frequently encountered.

    So yes, I agree with your intent, but I'd shift the focus to "most popular", which is actually much easier to measure than "best" (see the What is quality? discussion). I'd focus less on problem domain and more on hints for usage of each module. Actually the two don't have to be exclusive, an index by popularity and another by problem domain solves the problem. Note that as soon as you start sorting stuff into slots though, bits drop through the cracks - good indexes are very hard to generate! (The popularity index is easy - but mutable, the purpose index is hard - but fixed.)


    DWIM is Perl's answer to Gödel
Re: Section proposal: Best CPAN Modules
by tweetiepooh (Friar) on Jul 28, 2006 at 10:04 UTC
    It may be an idea to give a number of rating on modules such as ease of installation, number of dependencies, documentation and so forth.

    The fun I had building some modules that require lots of other modules which may require some library that is a real pig to build.

    This may be a better way of scoring a module than a simple score, so if there are 2 modules that do the same job but one is fairly simple, well doc'd and pure perl it may score more than the more complete, faster C module that can only be compiled with IBM's compiler on SCO Unix on a wet Friday afternoon provided the month has an 'M' in it. :)

Re: Section proposal: Best CPAN Modules
by fmerges (Chaplain) on Jul 28, 2006 at 12:52 UTC

    Hi,

    I think it would be interesting to have a list like this, because:

    • People want to solve a problem, not to search for 2 hours / module.
    • People still use modules for which there're better replacements/alternatives, because they're used to them.
    • People in doubt want to find others opinion or the main stream.

    People with no prior working experience in some fields, need a kind of guide, very often it's: use this module, and this.

    I do a lot of search.cpan.org looking at modules and also to the source code, to decide which module i'll use, and sometimes it's a tedious task.

    AnnotatedCPAN should be a kind of solution for that, but I would add also a rating system based on some criterias, to have something like the "hottest modules" which will vary depending on the module usage, develpment, etc

    And I don't think that the best way to produce better things is forking, or reinventing, things could be refactored; there is also CPAN Forum, RT CPAN, to speak about the development of modules, there's also email and so on, there could be more "community contributed modules". People actually do this, but there're a lot of silent consumers were we don't get feedback...

    Regards,

    fmerges at irc.freenode.net
Re: Section proposal: Best CPAN Modules
by mrguy123 (Hermit) on Aug 01, 2006 at 06:45 UTC
    I think that LWP::UserAgent is a great module. I have used it a lot for retrieving web pages and it gets the job done very well. SOAP::Lite is also a very strong tool, but I think it's documentation is missing a lot.
Re: Section proposal: Best CPAN Modules
by rvosa (Curate) on Aug 11, 2006 at 09:30 UTC
    I would be interested in usage statistics of modules - though I can't say for sure why it would be useful other than for spurious claims that popular equals good.

    A fairly quick way to get usage stats on more than just the modules that pop into everyone's head (classics like the DBI) would be if monks uploaded their perllocal.pod or autobundle to some website.

    (I personally would disable any 'call-home' scheme to gather such data.)

    You could even try to find clusters of modules commonly used together, in organically grown configurations ('webby', 'databasey', etc.)
Re: Section proposal: Best CPAN Modules
by kabeldag (Hermit) on Aug 16, 2006 at 09:11 UTC
    I think that having alot of modules available is great.
    You get a great variety of choice. The idea that anybody can write a Perl module and have it listed for the common good is a most beautiful feature of Perl and CPAN.

    I dont wish to rank modules by any category, neither would I want other modules to somehow become 'less-meaningfull' by some means. But, modules do need to be at least organized by purpose on CPAN. Perhaps more than they are now.
    This I suppose would at least make searching alot easier whilst still listing all modules and not leaving any out.
    Perhaps a search "modules by type" link on CPAN would be an assisting feature in module selection.

    Everyone creating modules should have an equal share in availability.

    I don't have an immediate answer. These things should be ground over by the community. I am for open source. Yet also for clarity in selection.

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: monkdiscuss [id://564134]
Approved by Paladin
Front-paged by Arunbear
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others pondering the Monastery: (19)
As of 2014-10-20 18:49 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    For retirement, I am banking on:










    Results (88 votes), past polls