unimatrix's Journal

Aspect-Oriented Programming in Perl

unimatrix — 2001-10-03T09:29:38+00:00

Aspect-oriented Programming comes to Perl.

I've uploaded an alpha distribution to CPAN, and it's also available from http://codewerk.unixbeard.net/aspects/Aspect-0.04.tar.gz. There is a "perl-aspects" mailing list at http://www.astray.com/mailman/listinfo/perl-aspects.

What is Aspect-oriented Programming?

Aspect-oriented Programming (AOP) is a programming methodology developed by Xerox PARC. The basic idea is that in complex class systems there are certain aspects or behaviors that cannot normally be expressed in a coherent, concise and precise way. One example of such aspects are design patterns, which combine various kinds of classes to produce a common type of behavior.

Another such aspect concerns debugging or tracing the flow of execution. Maybe every time one of a set of subroutines is entered or exited, you want to want to print a line telling you so. Normally you would you the good old 'print' statement debugger, or maybe you would actually use a custom Perl debugger or a C module. But what if the profile produces too much output you're not actually interested in? And a custom debugger usually does not make for reusable, easily customizable code.

You might change the source of all those subroutines you're interested in to print the desired information. When you are done debugging, you remove the print statements again. That's a lot of effort and can lead to bugs and oversights of its own. Furthermore, your code is littered with unsightly print statements. If you want to change the output, you have to change every print statement. If you add a class, remember to add the print statements. It's a mess. Wouldn't it be easier if you could influence the behavior of those subroutines from the outside, like a switch you could flip to make them print a statement when they're entered (in a consistent way, without chance of forgetting to create or remove a print statement), then flip the switch again to turn that behavior off?

That's what aspect-oriented programming is about. Aspects usually concern many classes right across the class hierarchy. One says that they "crosscut the program structure" and refers to them as "crosscutting concerns".

Just like object-oriented programming (OOP) encapsulates the concepts and functionality of individual classes and arranges them in a class hierarchy, so does AOP encapsulate the concerns of many classes and make those aspects or behvaviors reusable. In this way AOP is orthogonal to OOP.

Aspects allow you to change the interaction of objects without breaking their encapsulation. You don't change the behaviors of the objects per se, but rather the way they behave at the seams, at well-defined points during execution. For example, you can intercept method calls, override methods, influence operators, and generally influence the flow of execution in a way that isn't possible without AOP.

Certainly you can achieve all these things without AOP, but it would be a lot harder, less (or, more likely, not at all) reusable and more error-prone. Just as you can simulate any object-oriented program using non-OO techniques, but OO makes things just so much easier.

By modularizing such concerns into aspects you ensure consistent behavior across your project and don't depend on individual programmers to follow a convention, say, on how to write security or error checking. Indeed, it may not be necessary to for individual programmers to write code for such concerns at all, if they're centrally managed, from the outside, as it is, by aspects.

Exporter::Simple

unimatrix — 2001-07-03T12:07:01+00:00

Here is another module using attributes to simplify an API and hide implementation details. This time it's Exporter's API. Exporter::Simple, when used by a package, allows that package to define exports in a more concise way than using `Exporter'. Instead of having to worry what goes in `@EXPORT', `@EXPORT_OK' and `%EXPORT_TAGS', you can use two attributes to define exporter behavior:

my @bar : Exportable(vars) = (2, 3, 5, 7);
my $foo : Exported(vars) = 42;
my %baz : Exported = (a => 65, b => 66);

sub hello : Exported(greet,uk) { "hello there" }
sub askme : Exportable { "what you will" }
sub hi : Exportable(greet,us) { "hi there" }

You get the idea. I wasn't sure whether to emulate attributes for globals as well, but seeing that they're going to be in Perl 5.8.0, I'm going to wait until that's out before revising Exporter::Simple for a clean solution.

Enjoy. Feedback, as always, is welcome.

Perl 6 proof-of-concept page

unimatrix — 2001-06-29T14:41:13+00:00

I've done a web page with information and links to Perl 5 modules that may be of interest to those discussing and implementing the Perl 6 language. Some of them are proof-of-concepts of Perl 6 features that the respective authors have implemented to show how things might work in Perl 6.

It's at http://www.codewerk.com/perl6/.

Additions and suggestion for the page are welcome, please send them to marcel@codewerk.com.

Scalar::Properties

unimatrix — 2001-06-27T22:40:25+00:00

After reading a journal entry here, James Duncan sent me some pretty cool code showing how to get constant overloading and literals-as-objects working as intended. At that point, Scalar::Properties had become an inevitability. Let me amuse you with an excerpt of the module's synopsis:

use Scalar::Properties;
my $val = 0->true;
if ($val && $val == 0) {
print "yup, its true alright...\n";
}

my @text = (
'hello world'->greeting(1),
'forget it',
'hi there'->greeting(1),
);
print grep { $_->is_greeting } @text;

my $l = 'hello world'->length;

Graphing Makefiles

unimatrix — 2001-06-19T18:01:36+00:00

While playing with Makefiles, I had the urge to know what depends on what, hence GraphViz::Makefile was born. Coming to a CPAN near you rsn (i.e., after the most pressing other stuff gets done).

YAPC::NA afterglow

unimatrix — 2001-06-19T17:51:53+00:00

Revision: Without Damian, it would only have been half a conference, but without Kevin, Luc and Rich, there would have been no conference. Congratulations on organizing it and keeping it running smoothly.

On the flight back I was thinking about Ruby's literals-as-objects and realized that Perl can do this as well. By overloading constants, you can do things like

print 42->times(3);
print 'hello world'->reverse;

This could be used for runtime properties:

return 0->true;

Now we need to override the test for definedness to also check for this property along with the value.

YAPC::NA::END{}

unimatrix — 2001-06-19T17:01:18+00:00

YAPC::NA 2001 has finished. The time went by too fast, but it was fun. Lots of ideas and inspirations. And I've met many people whom so far I've only known from IRC or mail or newsgroups.

After the auction (where I actually bought a Ruby book and learned about its literal objects; nice) a bunch of us went for dinner in a lovely French restaurant. We split up into several groups as we didn't want to run the by-now-familiar risk of being turned away due to large numbers, and Abigail, Leon, Isabel (of NY.pm) and freeside had dinner, and we did something else afterwards, but I just can't seem to remember what. Drinking, probably. Someone mentioned to Abigail that he thought Klingon sounded like angry Dutch.

Saturday lunch we had the CPAN BOF, where we had to book a hotel room as there were about 30 people. Brian, who apparently hadn't slept all night, tried to convey his napc-idea (think CPAN + Napster for binary builds... watch this space) and there were other intense discussions. CPAN is great, and other languages would kill to have something like it, but there's still lots to do.

Then dha, Skud, Leon, Brian and many more went down into the old town to have a few drinks. A card game called "Chez Geek", the blame protocol and a long-expected rain (with impressive impromptu rivers down the sloped streets of Montreal) come to mind, in no particular order.

Sunday I just walked around lots; dinner at Mr Yan's again; now it's Monday and I'm writing this in a cafe just before going to the airport. Will be able to upload it only in Vienna, though.

NB: At the airport now. Complete rip-off: on top of an expensive flight, you have to pay CAD 10 when going through customs. "Oh, we just thought we'd rip you off some more before upholding our end of the deal."

YAPC::NA. Day 3. Lunch.

unimatrix — 2001-06-15T17:06:36+00:00

We had some problems with wireless this morning, so this entry is a bit late. We don't do steenkin' wires.

Yesterday afternoon saw Damian's amazing "Life, The Universe and Everything" talk - where he mixes Conway's (no relation) Life, Klingon, Perl 6, Maxwell's Demonsa and more in a talk that you just had to be there - Damian at his best.

This morning mjd gave his talk on the dirty innards of the regex engine and the identity function. Next was Leon's talk about creating compilers for little languages using Parse::RecDescent or Parse::Yapp and abstract syntax trees; a concept with huge potential. Very exciting stuff.

And finally I had my debut as a Perl conference speaker: A lightning talk on Aspect-Oriented Perl. I was a bit nervous just beforehand, but up on stage, it quickly passed. Even Abigail liked the idea. Bodes well for my first full-fledged talk on "Handling Attributes" at YAPC::Europe.

Speaking of stages, for the last lightning talk, Damian and Brian took the stage to present Inline::Files in a Shakespeare-style play, in which, amongst other things, Damian (or rather, his character) called Brian('s character) the "puppet of an active state". It was an incredible talk and there were standing ovations at the end.

A Perl conference without Damian is no Perl conference.

Now I'm looking forward to the Perl 6 status meeting after lunch. Speaking of which...

YAPC::NA. Day 2. Morning.

unimatrix — 2001-06-14T13:24:19+00:00

Fixed two bugs in Inline with Brian Ingerson. Discovered that Data::Denter is very slow with big data structures with circular references. Worked some more on Getopt::Attribute, which is being uploaded as we speak.

I survived Abigail's Parse::RecDescent talk.

Lots of talk, some more ideas, smoked meat bof (bof good, beef bad - but hey, at least they don't have foot & mouth here, so you gotta get all the meat you can). Then a few of us went to a french restaurant / bistro. Pretty nice.

Currently I'm in gnat's Perl Internals class. Tonight will still see Brian's 'Data Denter & YAML' talk, Abigail's 'Return of the JAPHi' and Damian's 'Life, The Universe and Everything'. As well as the YAPC dinner.

YAPC::NA Day 1. Morning.

unimatrix — 2001-06-13T11:42:19+00:00

Setting up registration in the afternoon was pretty painless . Went to a restaurant a few blocks away from the conference in the evening. Which turned out to be somewhat of an uncoordinated effort as poor Mr Yan couldn't cope with 40 hungry hackers, so about half had to be turned away as he "didn't want to give them bad service". Which is a nice attitude.

Met up with acme, Brian Ingerson, dha, Gurusamy Sarathy and had some more crazy ideas which this time I'm *not* going to add to my long list of projects as the list is already, well, long.

Now it's early morning (though the jetlag still caused me to wake up at 5.30am, so no change there) and they're just setting up breakfast and registration behind us as acme, gnat and me are updating our journals.

Some minor and not so minor glitches: Michel Rodriguez can't make it, so something else has to be arranged presto - for a 1.5 hours talk that's supposed to start at 10am. Fun!

gnat brought a wireless network base station - but no power cable. Thankfully there's an abundance of hubs to compensate. Nothing wireless, though.

Ah, lenzo has just turned up. Which means we'll have airport again soon.

Later.

yapc::europe talk

unimatrix — 2001-06-12T19:21:06+00:00

Well, my talk on "Handling Attributes" for YAPC::Europe has been approved. Which means some work after returning to Vienna after YAPC::NA.

live from yapc::na

unimatrix — 2001-06-12T19:19:09+00:00

This is unimatrix, reporting live from YAPC::NA over the wireless network which lenzo and moi (to use the local language) have just set up. Registration starts at 4pm (which is 43 minutes from now - and counting).

I am indeed in shorts as it's pretty warm. Yesterday we had a bit of rain which was compensated by humidity today.

The bad news is that Larry can't make it as he is apparently rather ill; Damian will do his opening talk instead.

A few guys have already arrived and are hanging around the registration area.

My lightning talk (aspect-oriented perl) coincides with acme's compiler talk, so I'll ask mjd to move my talk to the back. Trying to explain this concept in five minutes is madness anyway. The title should be "The Reduced Hacker Company: The Compleat Aspect-Oriented Perl (abridged)".

I'll send updates if and when I get around, although with wireless, it should be a lot easier.

Exporter::Simple, networking simple

unimatrix — 2001-06-07T07:42:10+00:00

Worked some more on Exporter::Simple, which uses attributes to make exporting variables (even lexicals!) and subroutines simple as you don't have to worry about Exporter's arrays and hashes.

Exporting lexicals was made possible/easy using Robin Houston's PadWalker module, which allows you to see which lexicals are defined in your scope.

Acme mentioned wireless networking in his journal; I have to agree that it's pretty cool wandering about with your powerbook downloading stuff - and no wires whatsoever. Let's see if we can get together a wireless network for YAPC::NA.

The lightning talk on aspect-oriented Perl has been approved, but it seems scary now to try and explain aspects in five minutes - apart from the fact that I've spent all time scheduled for aspects on Exporter::Simple... Oh well; it'll be a brief overview.

Entering ideas. Aspects.

unimatrix — 2001-06-01T14:40:47+00:00

Having a little notebook (of the paper variety) to jot down quick ideas is a good thing (I have a Palm but can never quite bring myself to use it); but every now and then the notes become so extensive that you have to start entering them lest you have to carry your laptop, notebook, additional loose pieces of paper, the Palm (that does get used from time to time) etc. Need I go on?

So it's typing time. Having it all on disk makes it easier to backup those ideas, but it's just no match for reading from real paper.

Coming soon to a Perl interpreter near you: I've had some more ideas for aspect-oriented perl; more on which at YAPC::North::America. Basically it's about being able to influence almost any point during execution from the outside, without having to rewrite (and un-rewrite later) loads of modules to install (and uninstall) certain behavior. Sounds abstract, I know. Read http://www.parc.xerox.com/csl/projects/aop/ and aspectj.org for details.

Attributes II

unimatrix — 2001-05-31T07:40:25+00:00

After doing some more experiments and work with attributes, there are now two more modules on CPAN: Attribute::Util, which provides four general-purpose attributes (Memoize, Abstract, Alias, SigHandler) and Attribute::Overload, which allows you to declare a subroutine to be an overload handler for one or more operations.

I've also submitted a proposal for a talk on attributes to YAPC::Europe (yes, I know the CFP deadline is today...)

Work still continues on Exporter::Simple, another module that uses - surprise! - attributes to make life easier if you want to export variables and subroutines. Damian is about to release a new version of Attribute::Handlers and so I'm delaying the release of Exporter::Simple; there are still a few things to be implemented.

I've quit my job (it was the end of a project) to spend all summer hacking Perl (ok, I'll try to get out every now and then...), and you just get so much more done if you can work a) on what you really love to do, b) in concentration, c) without commercial pressures.

I'll start looking for another job in September, but again it'll only be to fund some more time on what's really important.

Fun with Attributes

unimatrix — 2001-05-18T20:38:31+00:00

Got into attributes today. Damian's Attribute::Handlers really makes this easy. The result so far are the following three modules:

Attribute::TieClasses - attribute wrappers for CPAN Tie classes. The following lines load in Tie::Scalar::Timeout and tie()s $k with those options

use Attribute::TieClasses;
my $k : Timeout(EXPIRES => '+2s');

Attribute::Memoize - Attribute interface to Memoize.pm

use Attribute::Memoize;
sub fib :Memoize {
my $n = shift;
return $n if $n 2;
fib($n-1) + fib($n-2);
}

Attribute::Abstract - implementing abstract methods with attributes

package SomeObj;
use Attribute::Abstract;

sub new { ... }
sub write : Abstract;

Now, if properties (in Perl 6 jargon) were modifiable at runtime, it would open the door to aspect-oriented programming. There might be a little talk at YAPC::NA and/or YAPC::Europe on that; so far there is a proof-of-concept implementation of aspects using the flexibility afforded by the debugger. More on that later.

Why Perl?

unimatrix — 2001-04-09T15:29:28+00:00

I yearn for a sense of common purpose, with Open Source and
Perl hackers, beyond competing for the next job. We as
hackers are in the happy position of using and building
those tools that further sharing ideas. I'm optimistic that
the 21st century can be a blissful time of shared
conviction.

Why Perl?

Perl is not just a way of telling the computer what you
mean, it is also the basis of a thriving culture. The Perl
Mongers are one of the most important aspects of this
culture. They have chapters throughout the world, with new
ones being built all the time. Meetings, mailing lists, IRC
channels and grassroots conferences produce an energy and
exchange of ideas. This is helped by flat structures; even
the most brilliant minds within the Perl community are just
regular guys you don't have to be afraid to talk to.

Certainly one reason for this thriving culture is that fun
can be had with Perl, since the language is expressive and
flexible enough to allow you to write code the way you want,
be it obfuscated or poetic (the two are not necessarily
mutually exclusive), as a one-liner or as beautifully
structured object systems. There's more than one way to do
it.

Perl also allows you to do crazy things in easy and powerful
ways. Quantum-superpositioned variables that can be in
several states at the same time, or variables whose value
expires after some time; writing code in Latin or data
structures that flexibly bend and morph into any shape you
want, are but a few things possible in this rather natural
computer language - no contradiction in terms implied.

Elaborate code repositories, automated testing facilities,
recorded history; and now, the first development grant
sponsored by a community. Within two weeks of making public
the idea of sponsoring Damian Conway for a year to
exclusively work on Perl, the notion had become a reality.
Perl generates quite a bit of money, and good programmers
are sought-after, well-paid and tend to be generous and
helpful - as it has turned out, not only with advice. If
this concept works, it could spawn a wave of communities
self-sufficiently honoring brilliant minds. It is in this
environment that a sharing of ideas and common purpose is
fostered and achieved.

Open Source and Artistic Movements

unimatrix — 2001-04-09T15:28:37+00:00

It's easy to make plans when you're on holiday or have some
spare time, but once you're back home the day-to-day chores
will quickly catch up with you. Nevertheless, I think I've
learned something in the past year. Those two months in the
summer of 2000 that I've taken as more or less a sabbatical
have shown just how far you can get and how much fun it can
be when you're don't have to go to a regular nine-to-five
job. And experiences in London, meeting the open-source
community, namely in the form of the London Perl mongers and
taking part in the London Perl conference, have shown me
that, unless you're part of some culture you can relate to,
the life of a hacker can be a solitary one. However, when
you're engaged in open-source activities, you can immerse
yourself in this hacker culture, and it feels good and gives
you energy.

There are some concepts which may appear initially unrelated
to computer programming, but are compatible with that
mindset. The idea of living a bohemian lifestyle, of
exploring the artist within yourself (and hacking certainly
can be an art form), the comparison with the French painter
or poet around the fin-de-siecle who takes part in new
artistic movements, maybe even political discussions with
other artists in some Parisian cafe, drinking Absinthe (the
poet's third eye), is very appealing. In his talk at the
yapc::Europe 19100, Simon Cozens related the open-source
hacker to such artists. But how do you achieve such a state
of the free spirit?

The principal downside of programming for fun is the lack of
income. Only rarely can one find a job, temporary or
permanent, that allows oneself to indulge in open-source
projects. My rather temporary solution for this has been a
three-month visit to London, in a rather well-paid project.
London is expensive, Vienna is not, so the idea was to earn
money in London and spend it in Vienna. Back in Vienna, I'll
try to find similar set-ups; well-paid, short-term
commercial projects followed by periods of practising your
own hackerly art. Unfortunately, income taxation in Austria
is very high, so it would pay off a lot more working in
London, but then again that's not home.

There are, of course, other ways to make a living. Having
taught a few small Perl-classes, I found that to be
challenging and fun at the same time. So I might offer some
open Perl training with the Vienna.pm and Linux user group
folks, hoping to build up experience and a reputation, so I
can then offer these services to companies. Speaking of the
Vienna Perl mongers, they seem to be lazy beyond virtue when
it comes to organizing meetings and other activities, so I'd
like to encourage socal contact between those hackers.

Hello, World!

unimatrix — 2001-04-09T15:26:58+00:00

The first entry into a new diary.

There will be notes of a more general nature, not specific
to a particular project.