blogs.perl.org

Today, I've done something strange.

No, there weren't drugs involved, I merely sent a patch for Rakudo for a bug I reported a few weeks back. But the patch is... interesting.

First, about the "bug." Newest JSON spec lets you have anything as the top level thing. I spotted a few modules in the Perl 6 ecosystem that still expected an object or an array only, and the vendor-specific (possibly to be made hidden in the future) to-json subroutine provided by core Rakudo behaved the same as well.

One of the modules got fixed right away and today, seeing as there were no takers, I went in to fix the bug in Rakudo myself. Since I'm a lazy bum, I merely went in to that one fixed module and just copied the fix over!

But wait a second... ain't the Perl 6 module written in Perl 6? How did I manage to "just copy it over"? What sorcery is this! Surely you lie, good sir!

Last year I attended the Programming the web with Dancer master class at YAPC::NA, taught by the awesome folks Sawyer X and Mickey. It was then I learned that a new way to access get/post/route parameters was being baked and the potential pitfalls of what had been the current practice.

At the time, I was busy evangelizing the use of Dancer2 in professional work, and that meant exposing other developers at various levels of Perl familiarity how to accomplish tasks in the Dancer2 framework. While I waited for the new DSL parameter keywords be released by the Dancer2 core team, I decided to extend the Dancer2 DSL with keywords for parameter access of my own and added that to company broilerplate rather than have to worry about making sure every developer on every Dancer2 project was fully educated (or remembered) the aforementioned pitfalls when doing their work.

Haven't posted anything for a while, but I'm not dead, just busy. Here's a quick recap of things that I think people might find of interest.

First and foremost, I'm going to be in Brussels, Belgium, next weekend for FOSDEM. If you can make it, check out the Perl track. I'll be speaking about why people are finding Perl 6 so exciting. In particular, ever since the Christmas release, there's been a fair amount of chatter about Perl 6 and I've been paying a lot of attention to people who are looking at it for the first time, without a lot of preconceived notions. The reactions often range from "wow, that's cool", to "oh my goodness, I want that!" What's even more interesting is that they're not focusing on a particular feature (which would be scary as it would pigeonhole Perl 6 as a "niche" language). Instead, plenty of people getting excited about different things which scratch their particular programming fetishes: grammars, gradual typing, concurrency, and so on. It's fun to watch.

But there's more ...

https://codingkilledthecat.wordpress.com/2012/04/28/why-your-company-shouldnt-use-git-submodules/

With the gracious collaboration with Cosimo,Text::Hunspell made the switch from ExtUtils::PkgConfig to Alien::Hunspell late late year. Despite the somewhat complicated dependency requirements, this immediately made the spell checker more reliable as reported by cpantesters, as many machines do not have pkg-config or hunspell installed.

The one major platform that didn’t work on the initial switch was of course Strawberry Perl, but after some debugging and patches I got Alien::Hunspell and Text::Hunspell to work there as well. I even submitted patches to upstream to the hunspell project, which were accepted, so that in the future less patching will be required. This is what is great about Open Source when it works.

The results as recorded in the cpantesters matrix are stark:

• 2.14 - the first version to insist in its Makefile.PL on a version of Alien::Hunspell that works on Windows
• 2.11 - the last version to use ExtUtils::PkgConfig

I'm still trying to get my head around how the concurrency stuff works. I had an idea for a simple script: sometimes I run a program that dumps a lot of output, so I'd like a script that would take that as input and just hold onto the most recent X lines of it, so that they could be read at any point through a named pipe.

So I figured I'd need one thread to read lines from STDIN ($*IN in Perl 6) and keep an array loaded with a certain number of the most recent ones, and another thread to open the fifo for writing and dump the current contents of the array to it whenever someone reads it. In Perl 5, I'd probably do it by forking off a child process for half of the work, or maybe use a single loop that does both things but has quite a few no-ops. I figure in Perl 6 I can do it with threads.

As a Perl 5 programmer of about a decade, I'm well aware of how it was referred to at some point or another as the "write-only" and "linenoise" language. With the newest addition of the baby Perl 6 language to the Perl family, I fear that I must declare (wildly speculate) based on my extensive research (a boring ride on a bus, while staring at my phone) that Python steals that title now!!

Why Python? Blame whoever made the Stackoverflow Python Report scroll through my Twitter feed. I merely picked two problems from that list and asked myself what would the Perl 6 solutions to them look like.

Interleave Two Strings

The top rated item of the week is interleaving two strings.

#Given:
u = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'
l = 'abcdefghijklmnopqrstuvwxyz'

#Wanted:
'AaBbCcDdEeFfGgHhIiJjKkLlMmNnOoPpQqRrSsTtUuVvWwXxYyZz'

The accepted Python answer is this:

res = "".join(i + j for i, j in zip(u, l))
print(res)

I am creating Validator::Custom from 2009. This is a validation module.

I add many features to customize validation but I can't solve many problems because Validation logic is more complex than I expect.

Mojolicious remove filter feature of validation module. It is maybe impossible to do three things(require, checking, filtering) at once.

And validation logic is complex because we want to see multiple values at once. for example, "high_price" param + "low_price" param.

Next, In html validation, we must think about fore cases.

1. name=foo
2. name=foo&name=bar
3. name=
4. (not exists)

And We want do filtering for blank containing value, for example, " Kimoto ".

And We maybe want to prepare default value when validation fail.

Above combination is very very complex.

Conclusion is that simple is best

At first programmer must to choice a value or values. By this, we don't need to thing "require" checking.

  # A value
  my $age = $c->param('age');

  # Values
  my $features = $c->every_param('features');

After that, you do filtering and checking.

  $age = $vc->filter($age, 'trim'); # "  12 " to "12"
  my $age_is_valid = $vc->check($age, 'int');

  $features = $vc->filter($features, 'remove_blank'); # ['', '001', '002'] to ['001', '002'] 
  my $features_is_valid = $vc->check_each($features, 'in', ['001', '002', '003']]);