DEV Community: Tib

My participation to the Perl Toolchain Summit 2025

Tib — Thu, 27 Mar 2025 07:36:14 +0000

What is the Perl Toolchain Summit?

The Perl Toolchain Summit (PTS) nee "Perl QA Hackathon" is an amazing event, in the form of an hackathon, taking place yearly in Europe.

It's a open source gathering of several critically important people running the Perl infrastructure.

(picture from Ant Rozetsky)

Some of the vital pieces of software taken care at Perl Toolchain Summit are:

PAUSE, CPAN and MetaCPAN
CPAN installers: cpan, cpanp, cpanm, cpm, etc...
Builders: ExtUtils::MakeMaker, Module::Build, Module::Build::Tiny
Smokers: CPANTesters, Test::Smoke
Authoring tools: Dist::Zilla, Dist::Milla, Minilla, etc...
Test frameworks: Test2, etc...
CPAN Security

This gathering, invite-only, is happening once a year and group together a lot of essential folks running Perl and CPAN.

It is made possible thanks to the generous sponsoring of companies.

By sponsoring this event, companies both mark their strong belief in Free Software and help maintaining the present/future quality of Perl ecosystem (or give thanks for past good and loyal service for years).

It radiates well beyond the event since the attendees are committed to Perl all along the year before and after the summit.

To give you an example, see the list of achievements of CPAN Security group in 2024 (which is particularly relevant since CPANSec group was created at Perl Toolchain Summit 2023 and will be again well represented during this summit)

What about /me?

This year, I got invited to the Perl Toolchain Summit. It will be my first time :)

I felt honored and blessed to receive an invitation.

"Woohoo!" \o/

I'm now feeling excited to join this event and meet all those great names.

I'm far from the value of some core members participating to the Perl Toolchain Summit but I think this invitation rewards my involvement in Perl toolchain (and Perl in general). So I would lie to say that I'm an imposter. But still, I will try be up to the chance I'm given!

My plan

I'm involved in multiple topics but I'm thinking I will work mainly on the 3 following topics:

Perl interpreter smoke test tooling (for which I'm the maintainer now)
Security (along with CPANSec group) where I've been involved lately
CPAN in general (PAUSE, indexes, installers, architecture, etc...)

Topics 2. and 3. are largely overlapping themself.

For the topics involving collaboration, I'm not going with big breaking ideas but mostly with little realistic changes that I consider with relatively big "return on invest". I think there's some inertia (because of the compatibility value being deep rooted in the DNA of Perl) that we should respect and should make us "reasonable dreamers".

If you're used to read me, you know that I'm also thinking out of the technical box. Sometimes speaking about the social nature of CPAN or how communication impacts as much as technical excellence.

I'm not fixing myself "success criteria" to fulfill to consider I would have succeeded my summit. If I'm preparing a bit, I also want to keep some "freshness" toward the event.

What else to say now apart that I can't wait to be there!

Meanwhile...

Happy Perl Hacking!

Habemus Perl Logo!

Tib — Sat, 04 Jan 2025 16:25:36 +0000

The new Perl logo

Perl has a logo!

I mean, a FREE logo.

It's a camel, with good quality and not a derivative of O'Reilly camel...

Take your time to fall in love with it.

Use it... or not, no obligation.

But we have one, a free one, that we can use.

References

Please read Perl Advent Day 23 2024 - A New Logo for Perl
Find source images for usage at perl-assets

Further on Perl Camel

Tib — Thu, 07 Nov 2024 09:07:12 +0000

Get enough of Perl logo discussion?

You would well take a little bit more! 😉

(Picture from Askar Abayev)

I know some people think it's not worth the pain discussing again, but I suppose it's because they don't have the same needs than me and some others 😀

What if you want to use a logo to illustrate a Perl blog? What if you want to use an icon to illustrate Perl sources files (Ubuntu)? What if...?
You either have to use a random camel or ignore the law...

And it's in Perl roots and inside us to be good at free software.

If you don't know well the context, please go read my post The complex topic of Perl logos

Later, I posted my Open Letter To Tim O'Reilly To Free The Perl Camel on Hacker News and people mentioned about the fact that Perl Camel was a "trademark" of O'Reilly, not a copyright (license).

It is true and it's a fundamental thing to understand.

What's the difference between trademark and copyright

Trademark vs Patent vs Copyright

How is trademarked Perl Camel?

First, just having used the Perl Camel on a book provided to O'Reilly Media a "public statement", then quickly after (in 1998) they secured the trademark via registering trademarks against organisms like United States Patent and Trademark Office (USPTO).

Public statement?

Once a business begins to use a logo to advertise and sell goods or services, they automatically attain common law ownership and the limited protections associated with it—but only if the business can prove it was the first businesses to use the mark. Common law legal protection for a trademark only applies to the geographic area where the business operates. If nationwide protection is needed, then registering a logo with the USPTO is a wise step.

USPTO / WIPO

United States Patent and Trademark Office (USPTO) and World Intellectual Property Organization (WIPO) are organisms where one can register a trademark.

Trademarks are not "global", they apply only in jurisdictions (countries) who recognize it.

We can find USPTO registrations for Perl Camel, I was not able to find one in WIPO (is there any?).

The actual trademark and service mark

"The bottom line is that a trademark represents a product produced by a business (in this case: The Programming Perl book) while a service mark represents a service offered by the business (in this case: O'Reilly selling books). A single business can register both trademarks and service marks to brand themselves."

Trademark

Serial: 75414570
Reg number: 2262465
2nd renewal: 20190828

Camels, both bactrian and dromedary, including heads and other parts uniquely identifiable with camels, and also including stylized, costumed, and those with human attributes

Service Mark

Serial number: 75414568
Reg number: 2209120
2nd renewal: 20181012

You can search and verify by yourself at USPTO Trademark Search System

Trade dress

More in general, covers of animals (camels but also others) on O'Reilly books seem protected under a Trade dress:

"Trade dress is the characteristics of the visual appearance of a product or its packaging (or even the design of a building) that signify the source of the product to consumers. Trade dress is an aspect of trademark law, which is a form of intellectual property protection law.

In the United States, the Lanham Act protects trade dress if it serves the same source-identifying function as a trademark. It is possible to register trade dress as a trademark, but for practical reasons most trade dress and product configurations are protected without registration."

If copyrights have a duration, how long lasts a trademark?

10 years but you can renew (O'Reilly well renews them when needed).

What is the copyright (license) of O'Reilly Perl camel?

Actually, the design was free when O'Reilly decided to use it.

O'Reilly animal designs are copyright free (public domain). It implies it has no restriction and no license.

Edie Freedman and Hanna Dyers designed the Programming Perl cover based on Animals 1,419 Copyright Free Illustrations of Mammals, Birds, Fish, Insects, etc. by Jim Harter

So it's important to understand that Perl Camel is not protected because of any copyright/license or creator but because of its association with O'Reilly and books publishing.

Using Perl shadow camel?

Yes, what about this guy?

Being a derivative of the Perl Camel does not free us from the trademark. Like using Coca Cola logo just replacing with a "Koka Cola" or changing color from red to blue. No derivative can help, design should not recall in any way the initial trademark/picture and, moreover, the association.

Fair use

The concept of fair use refers to some usages of trademarked logo for some limited usage like representing the brand. For instance, using the Perl Camel to write a blog post about the Perl Camel Book or O'Reilly in general is OK.

Fair use is not well recognized everywhere though.

Blanket license

It exists some solutions to delegate permissions, in the sort of "automatic grant of permissions to use". Somewhat like a blanket license. What was done with The Perl Foundation looks like a bit to a blanket license, but with restrictions. The way it was done makes it still very far from making it "free to use".

Expanding blanket license (remove restrictions and do not attach to TPF) is difficult or impossible without an entity (Perl "community" is not what we can call "an entity").

Logo vs mascot

Using camels to represent Perl has become more like a mascot.

My hypothesis and conclusion

I'm not a lawyer 😉

When we talk about trademarks, we don't talk about something really rigorous. For instance it works on some countries, but not all.
When defending a trademark in court, it's not all black and white, it's more like you build a defense with a maximum of documents and history facts to prove your precedence and that you undoubtly own the trademark (and trade dress).

My hypothesis is that O'Reilly never trademarked other animals with books in his entirety but Perl Camel was the first one (to be trademarked, not the first published). Protecting books covers with animals seems to fall under a "Trade dress" protection, built on top of the Perl camel trademark registration (among maybe other registrations). It means Perl Camel trademark can now be used to protect all others. Sadly for us, it is probably then the cornerstone of the whole protection of "O'Reilly Trademark family" (or more appropriately: "menagerie").

It's why if we dream to have it free, we can't reasonably expect to see it happen. And even if O’Reilly would lose total interest in Perl, it would weaken too much their whole family of books.... And they still remain friends that we don't want to hurt too much.

My London Perl And Raku Workshop 2024

Tib — Tue, 29 Oct 2024 14:38:19 +0000

Few days ago, I participated to London Perl And Raku Workshop and I was shocked and thrilled to see how it was a brilliant success!

For once, I will say a word about Deriv which was the biggest sponsor of the event.

It's a nice place to work 👍 for both juniors and gray hairs (that are very much desired). There's an office in Reading!

Also for once, I will post a picture of me.

The review

The venue was great. Both the location, size and equipment.

When arriving, you got a sticker to wear with your name (great idea!) and a gorgeous scarf (best goodie ever!).

The venue was literally crowded of Perl developers. With a ratio of 2/3 Britishers and 1/3 of foreigners.

It started with a keynote from Julien:

I attended the very much awaited talk from Paul Evans, on Perl syntax, CPAN experiments and Perl future:

In front of this talk was the another very much awaited talk "Perl Diver" that I had to miss (but I managed to steal a picture from it).

Then it continued with a Raku talk ("RakuDoc v2").

I liked this talk.

Then the usual "TPF updates" with half of presentation and half of questions.

One a side (TPF) note, Makoto Nozaki took a recent great initiative with a 3 parts blog posts, doing a big gift to the foundation and everyone, be sure to read them:

This transparency is needed for solid trust in our foundation. It's very much welcome and breaks some relative opacity (not on purpose) of the foundation. So once again thank you Makoto!

Then, a talk that I would never miss, OpenTelemetry from José Joaquín (JJ) Atria. Never deceived by a talk from JJ Atria!

The picture does not render it, but small room was so full that organizers had to refuse people to enter (but by luck I arrived earlier).

And it was already half of the day track...

Went at hotel for lunch and talk preparations. My talk rehearsals took longer than expected and I missed first talks of the afternoon 😢 and also missed the CPAN Security BoF 😢 that was in parallel.

Still was able to attend a later talk about security (Meterian)

And take a moment in a room to work closer with Sebastien, Olaf And Leo (in the common space).

And finally we reached the lightning talks, maybe the activity that I enjoy the most in conferences. I delivered a little talk ("This is a camel") and listened to others (Paul about sdview, Tux about DBI, Dave about Perl School etc...)

Also, Theo Van Hoesel did a tribute to Abe Timmerman, it was very poignant when the whole room did a minute of silence.

The conference ended with a keynote from Lee Jo, revealing the undercover of such event organization.

Visiting around the venue

The venue location was nice.

I walked around it, first reaching Bunhill Fiels cemetery where I relaxed for a while. The place, even being a cemetery, is more like a little garden, and it reigns a absolute calm, it's very pleasant.

William Blake has its tomb here.

I went to the south to reach Saint Paul's Church. My first impression was that it is gigantic (and beautiful).

Another very beautiful place is Guildhall library.

Totally by luck, I discovered a Banksy fresque (the "Banksy Basquiat") in Golden Lane, when going back to hotel after pub on Friday!

Before the event, we enjoyed a beer at Craft Beer (45 min from venue, just good for me). A typical pub with a large choice of beers.

After the event, I even more enjoyed the event at "The Eagle" pub (more people and hello 👋 Stig, Rolf, Salve, Robert...)

Fun

The accent problem

It's not a surprise but in UK, as a French guy, and more of that from South of France, I have an accent problem.
It means people don't catch some of my words, even some simplistic ones like "London", "City Road" or "Burden". And worst, my nickname "tib".
Second, it means that my comprehension is very reduced because of British accent (and native using more vocabulary).

UK is same but different

First UK is not anymore in EU, so the need of passport.
Second, UK is geographically aligned to where I live, but not aligned as a timezone.
Third, of course I forgot about the different power supply format and had to buy an adapter.

Cultural shift

London, at least the area where I stayed, was full of good energy.
You really feel the "go to pub for a beer" cultural habit (when we do "apero" in France), and my hotel had installed a fixed beer bottle opener on a wall.

Errol Court / Error Count

Do you like me read "Error Count"?

Too long time spent as a programmer...

Red Hat

I finally found where we can buy ~~Red~~ Fedora Hats!

Free library

A very picturesque free library

Perl Monks Pub

Is it where Perl Monks gathers at end of day?

Conclusion

Going to such even was such a lot of fun, socializing and charging battery that I'm more than grateful to have been part of it. I really feel like "privileged".

Again, it was a a brilliant success, thanks to organizers, speakers, attendees and all.

See you again and "Long live to Perl!"

App::cpx

Tib — Thu, 05 Sep 2024 17:56:04 +0000

(Picture from Erda Estremera)

Background

I'm sometimes doing Front End dev.

Or sometimes the best tool for the job is only installable via npm.

It can be scripts to "uglify" or "beautify" css/js, optimize svg files (svgo) or clients to SaaS platforms (wrangler).

Actually, it's not that important if it's part of javascript ecosystem, what I want is just to execute them!

The usual process: I start testing it locally, quickly and in a "trash-able" way, then install in a continuous integration pipeline, then forget 😄

So far, do you feel some resemblance with some periodical process or yours?

For this purpose, I'm using frequently npx (now part of npm).

Do you see now where I'm going to? 🤔

App::cpx, a `npx`-like for CPAN

I just uploaded App::cpx for this purpose.

Give cpx a binary and it will find it in CPAN, install it for you then execute it.

$ cpx hr -s 40
🎯 Found [bin/hr]
📦 Release to install [https://cpan.metacpan.org/authors/id/W/WO/WOLDRICH/App-term-hr-0.11.tar.gz]
🔧 Will install into /home/tib/cpx-test/.cpx
DONE install Term-ExtendedColor-0.504
DONE install App-term-hr-0.11
2 distributions installed.
=======================================

(purpose of hr is to draw horizontal lines)

Or another example with mlocate:

$  cpx mlocate Redis Moo
🎯 Found [bin/mlocate]
📦 Release to install [https://cpan.metacpan.org/authors/id/C/CE/CELOGEEK/App-Module-Locate-0.7.tar.gz]
🔧 Will install into /home/tib/cpx-test/.cpx
DONE install Module-Locate-1.80
DONE install Module-Build-0.4234
DONE install App-Module-Locate-0.7
3 distributions installed.
/usr/local/share/perl/5.34.0/Redis.pm
/usr/local/share/perl/5.34.0/Moo.pm

mlocate is a script that lives in App::Module::Locate. It's an utility to "find a module by its name".

cpx saves you from some frustrating tries:

Is it cpm install App::mlocate?
OK try capitalize cpm install App::Mlocate, I know some authors do that
Hmm maybe simply locate like this cpm install App::locate
Oh wait, capitalize cpm install App::Locate
Raaah! Stop it and open MetaCPAN and search
OK it's cpm install App::Module::Locate but now it's installed locally, how do I set local::lib? Maybe I should install globally
Hmm I will install globally even if it can be a bit dirty for a one time use

And at start were you 100% sure mlocate was something that exists in CPAN? 🤔

cpx saves you from this pain and hides you the internals of installing the module, if missing.

When running cpx again, it won't reinstall but reuse the already installed binary:

$ cpx mlocate Redis Moo
⚓ Found executable already installed
/usr/local/share/perl/5.34.0/Redis.pm
/usr/local/share/perl/5.34.0/Moo.pm

Your companion for Continuous Integration

$ curl -sL https://git.io/cpm | sudo perl - install -g App::cpx
$ cpx hr -s 40

In GitHub Actions, it would give something like this:

name : Test cpx
on: push
jobs:
  cpx:
    runs-on: ubuntu-latest

    steps:
      - name: Install cpx
        run: curl -sL https://git.io/cpm | sudo perl - install -g App::cpx
      - name: cpx hr 
        run: cpx hr -s 40

And see the result:

Conclusion

As of now, the code for this small utility is ridiculously simple (source), but sometimes good ideas (yes, all glory to myself 😀) are simple to implement.

Resources

I'm still lazy - Great module discussed here and I stole the idea of using App::cpm::CLI

SlapBirdAPM with Mojolicious

Tib — Tue, 27 Aug 2024 11:14:25 +0000

This is my quick walk-through exploration of SlapBirdAPM with my Perl web framework of choice (Mojolicious of course!)

Setup Mojolicious Web App

Install Mojolicious:

$ cpanm Mojolicious
--> Working on Mojolicious
Fetching http://www.cpan.org/authors/id/S/SR/SRI/Mojolicious-9.38.tar.gz ... OK
Configuring Mojolicious-9.38 ... OK
Building and testing Mojolicious-9.38 ... OK
Successfully installed Mojolicious-9.38 (upgraded from 9.32)
1 distribution installed

Generate project:

$ mkdir webapp
$ cd webapp
$ mojo generate lite-app bird
  [exist] /home/tib/Code/webapp
  [write] /home/tib/Code/webapp/bird
  [chmod] /home/tib/Code/webapp/bird 744

Test:

$ morbo bird
Web application available at http://127.0.0.1:3000

And check in your browser http://127.0.0.1:3000/

Create SlapBirdAPM account

Go visit SlapBirdAPM:

Create new app:

And get your API Key:

Install SlapBirdAPM in Web App

First install Mojolicious plugin:

$ cpanm SlapbirdAPM::Agent::Mojo
--> Working on SlapbirdAPM::Agent::Mojo
Fetching http://www.cpan.org/authors/id/R/RA/RAWLEYFOW/SlapbirdAPM-Agent-Mojo-0.005.tar.gz ... OK
Configuring SlapbirdAPM-Agent-Mojo-0.005 ... OK
==> Found dependencies: Const::Fast, System::Info
--> Working on Const::Fast
Fetching http://www.cpan.org/authors/id/L/LE/LEONT/Const-Fast-0.014.tar.gz ... OK
Configuring Const-Fast-0.014 ... OK
Building and testing Const-Fast-0.014 ... OK
Successfully installed Const-Fast-0.014
--> Working on System::Info
Fetching http://www.cpan.org/authors/id/H/HM/HMBRAND/System-Info-0.065.tgz ... OK
Configuring System-Info-0.065 ... OK
Building and testing System-Info-0.065 ... OK
Successfully installed System-Info-0.065
Building and testing SlapbirdAPM-Agent-Mojo-0.005 ... OK
Successfully installed SlapbirdAPM-Agent-Mojo-0.005
3 distributions installed

And add plugin SlapbirdAPM; to your web app.

Our full application now looks like this:

#!/usr/bin/env perl
use Mojolicious::Lite -signatures;

plugin 'SlapbirdAPM';

get '/' => sub ($c) {
  $c->render(template => 'index');
};

app->start;
__DATA__

@@ index.html.ep
% layout 'default';
% title 'Welcome';
<h1>Welcome to the Mojolicious real-time web framework!</h1>

@@ layouts/default.html.ep
<!DOCTYPE html>
<html>
  <head><title><%= title %></title></head>
  <body><%= content %></body>
</html>

Test!

Let's restart the application with the plugin and an API Key:

$ export SLAPBIRDAPM_API_KEY=01ABCDEFGHIJKLMNOPKtest-app && morbo bird
Web application available at http://127.0.0.1:3000
[2024-08-27 11:33:27.51107] [8402] [info] Slapbird configured and active on this application.

Use your favorite browser to browse http://127.0.0.1:3000 or curl http://127.0.0.1:3000.

Your activity will immediately reflect in your SlapBirdAPM portal

We're done! 😀

3 original Perl CPAN social concepts

Tib — Fri, 23 Aug 2024 11:00:30 +0000

(Picture from Eren Li)

CPAN Lieutenants

(Picture from Alex Andrews)

I often mentally represent modules "per namespace" (I explained it here) like in a tree where we have a node "JSON" that contains all JSON related (JSON:PP, JSON:XS, etc...).

I know it's partially wrong, there's no such "order" and for instance Cpanel::JSON::XS is out of this hierarchy but still well a JSON thing. But most of the time, this mental model works.

If we care about the quality of the Perl ecosystem, we should try to find a solution to fight the CPAN inertia.

Indeed, what happen when a module is broken and maintainer is inactive/unresponsive?

Some important modules can be considered a bit a "common good" (is it wrong to think like this?) but we need also to respect the ownership and copyright.

So here is the concept of CPAN Lieutenants.

For some very specific cases (e.g. a module does not install at all), I think we should be able to "force apply" a patch to a module.
Boundaries should be clear:

Purpose is to fix a broken module
Fix is relatively simple (to review), blocker is about releasing it

In practice, it would mean to let few trusted members to manage a dedicated sub-tree of CPAN namespaces. And in extreme case, after some formal review, discussion and notifications, these people would be able to push a fix to CPAN (= publish a new release).

It would be a bit like these people, thanks to their trust and expertise, have implicit co-maintainer rights (CO-MAINT in CPAN jargon) on a topic (e.g. "Alien" modules, or "XML").

It won't imply these "lieutenants" to maintain these modules neither the ceremonial of asking (and eventually being refused) co-maintainer permissions. And don't hear what I did not say: I do not criticize PAUSE operating model for that.

The name is inspired from Linux kernel development with "lieutenants" that have responsibility on sub parts of the kernel.

Also we need to respect authors and their rights. It's a matter of finding the right balance so everybody wins. That's why we should be careful with boundaries and rules (e.g. a minimum number of lieutenants approvals), the same way PAUSE admins are careful with module ownership.

If it proves to be incompatible with license, we could consider a layer of patches, inspired by what do several Linux distributions (see for instance Locally applied patches: section under Debian/Ubuntu perl -V).

It proved to work (for inertia and licensing) but I would prefer to avoid this, first because it's generally considered good practice "to push patch upstream" when possible, also because it would add complexity and "yet another layer of patches":

CPAN release contains base delivery
GitHub locally applied patches (merged commits in GitHub that never reached CPAN)
CPAN locally applied patches ⚠️
OS Distribution locally applied patches

But on the other hand, the CPAN local patch layer approach has advantage of managing the storage of patch (since inertia very likely also concerns code forge...)

I don’t know the exact solution, but I think there’s something to do 😄

Actually, a recent discussion on perl development mailing list ("p5p") has highlighted the same need and discussed some close proposals. I swear I did not copied! 😀 But it decided me to post this!

edit 1: a CPAN module security vulnerability is also a valid reason to want this and I see that CPANSec folks have similar concern, see CPAN Security Patch Tooling

edit 2: A Social Contract about Artistic Control - perl core statement (applies to modules added to perl core distribution)

edit 3: DistroPrefs - CPAN.pm patching facility

CPAN Mentors

(Picture from Thirdman)

I think the barrier of entry of Perl is high but with a compatible mindset, once someone get in, Perl (and CPAN) produces often long time involved members.

These days, there's no so much new entrants to the CPAN.

We should optimize for retention (and unity).

The idea of CPAN Mentors is to offer help to newcomers, once they pass the entry door, something that we can monitor from perl modules mailing list

Let’s say after 3 weeks, if a new CPAN AUTHOR has not published any module, someone can drop a little mail:

Hello new CPAN AUTHOR,

I noticed you recently registered to PAUSE/CPAN, congratulation for this.
I also noticed that you did not publish any module, yet.
I don't know your level of expertise,
maybe you would benefit from some guidance?
I'm reaching you to kindly offer my help :) 
Feel free to ask me questions or we can maybe even accommodate some pairing via a call.

Regards.

Tib

I think the number of new monthly CPAN registrations would permit it, and reinforce the community feeling that always been a strength of Perl.

CPAN Guards

(Picture from PhotoMix Company)

This concept joins actually a bit the CPAN Lieutenants concept.

The idea is to watch (and contribute to) a limited set of critical modules (depended by several modules or in Perl jargon: those high in the CPAN river) and help keeping them in good quality and portable.

This is actually not a new concept (forgive me the "original" in title), some people (CPANTesters and perl core developers) are already on it (see for instance BBC and CPAN River 5000).

So it's more about formalizing this under a fancy name.

CPAN Guards concept fits well the Perl folks that are more contributors than maintainers and prefer to jump often from modules to modules.

At some point it's also related to the Perl Phalanx initiative (more focused on tests)

Some reading and watching

The perl.fish experiment

Tib — Thu, 22 Aug 2024 08:06:37 +0000

(Picture from Chokniti Khongchum)

I like GitHub Pages, it provides a nice little and simple free hosting. I also like their static site generators (SSG) companions:

Hugo - used by example for perl.com (source)
Jekyll - that I used for The ephemeral miniconf (source)

About 3 years ago, I spent some time experimenting with more advanced Hugo and Jekyll concepts (variables, config, layouts, includes, ...).

I wanted to produce a modular website with content separated from a complex structure, using parameterized includes.

The experiment also contained a commenting system powered by GitHub issues (yes!) along with a blog (it’s Jekyll after all!).
I used this effort to compare Hugo and Jekyll features (Jekyll won for me).
Then I continued by adding some content, syntax highlighting and various other things...

But why I tagged this post with #perl tag?

Because ultimately the website content is about Perl, with some generic informations and links that I tried to present in a bit "catchy" way.

After some time, I let this website aging somewhere in my GitHub private space.

Nobody asked for it, but this is finally it!

🐟 perl.fish 🐠

It’s public but I have zero plan for this.

And about the name "perl-dot-fish", I doubt I can give any explanation except that I consider that "it sounds good".

A little preview

For the very most lazy 😀, here is a little preview:

It’s also looking good on mobile.

Note: for the design skeleton, I started from rust-lang.org (license permits it), but it now has well diverged.

About the stack

Registry: Identity Digital for the .fish (I wanted to pick a not so common nTLD)
Registrar: OVH
DNS: OVH
Hosting: GitHub Pages
Framework: Jekyll (SSG)

I built using Jekyll parameterized includes, to me it makes it both modular and a bit inelegant 😀
Website is easy to edit and the GitHub Pages setup deploys on commit.

Read the source at perlfish

It’s and experiment, probably running for about a year.

🐟 perl.fish 🐠

Installing CPAN modules from git

Tib — Tue, 21 May 2024 18:18:32 +0000

(picture from elevate)

For various reasons, you might want to install CPAN modules from a git repository.

It can be because somehow a git repository is in advance against CPAN:

A fix was merged in official git repository but never released to CPAN
A branch or a fork contains some valuable changes (this very-little-but-absolutely-needed fix)

Or it can be because the modules are actually not in CPAN: not public and not in a alternative/private CPAN (see Addendum) or simply they are only "experiments"

But this post is not meant to discuss about the "why" but instead mainly share technically the "how" you could do that 😀

I tested various syntax and installers and will share now some working examples.

☝️ Before we continue, be sure to upgrade your installers (App::cpm and App::cpanminus) to their latest

Installing from command line with `cpm`

Installing with cpm is straighforward:

$ cpm install https://github.com/plack/Plack.git --verbose
33257 DONE fetch     (0.971sec) https://github.com/plack/Plack.git
33257 DONE configure (0.033sec) https://github.com/plack/Plack.git
33257 DONE resolve   (0.031sec) Clone -> Clone-0.46 (from MetaDB)
...
33257 DONE install   (0.364sec) URI-5.28
33257 DONE install   (0.046sec) https://github.com/plack/Plack.git
31 distributions installed.

It can also work the same with ssh git@github.com:plack/Plack.git:

$ cpm install git@github.com:plack/Plack.git --verbose
64383 DONE fetch     (2.498sec) git@github.com:plack/Plack.git
64383 DONE configure (0.039sec) git@github.com:plack/Plack.git
...
64383 DONE install   (0.045sec) git@github.com:plack/Plack.git
31 distributions installed.

Installing from command line with `cpanminus`

Installing with cpanm is not harder:

$ cpanm https://github.com/plack/Plack.git
Cloning https://github.com/plack/Plack.git ... OK
--> Working on https://github.com/plack/Plack.git
...
Building and testing Plack-1.0051 ... OK
Successfully installed Plack-1.0051
45 distributions installed

Installing from `cpanfile`

The correct syntax is the following (thank you @haarg):

requires 'Plack', git => 'https://github.com/plack/Plack.git', ref => 'master';

(ref => 'master' is optional)

And it would just work later with cpm:

$ cpm install --verbose
Loading requirements from cpanfile...
33257 DONE fetch     (0.971sec) https://github.com/plack/Plack.git
33257 DONE configure (0.033sec) https://github.com/plack/Plack.git
33257 DONE resolve   (0.031sec) Clone -> Clone-0.46 (from MetaDB)
...
33257 DONE install   (0.364sec) URI-5.28
33257 DONE install   (0.046sec) https://github.com/plack/Plack.git
31 distributions installed.

⚠️ Despite being a cpanfile, please note the use of cpm

Installing from `cpmfile`

Let's write our first cpmfile and save it as cpm.yml:

prereqs:
  runtime:
    requires:
      Plack:
        git: https://github.com/plack/Plack.git
        ref: master

And then it would just work with cpm:

$ cpm install --verbose
Loading requirements from cpm.yml...
66419 DONE resolve   (0.000sec) Plack -> https://github.com/plack/Plack.git@master (from Custom)
66419 DONE fetch     (1.695sec) https://github.com/plack/Plack.git
66419 DONE configure (0.034sec) https://github.com/plack/Plack.git
...
66419 DONE install   (0.023sec) https://github.com/plack/Plack.git
31 distributions installed.

Beware of "incomplete" repositories

Releases on CPAN are standardized and generally contain what is needed for installers, but distributions living in git repositories are more for development and very often not in a "ready to install" state.

(thank you @karenetheridge)

There's some limitations that you can encounter:

cpm would refuse to install if no META file is found (but cpanm would be OK with that)
cpm would refuse to install if no Makefile.PL nor Build.PL is found, except if x_static_install: 1 is declared in META (cpanm would still refuse)

Should I mention the repositories with only a dist.ini? (used by authors to generate everything else)

And you would get similar trouble with distributions using Module::Install but having not versioned it.

Conclusion

You should probably not rely too much on "install from git" method but still, it can provide an handy way to install modules to test fixes or experiments.

And now with this post you should have good examples of “how” you can achieve that.

Addendum

Alternative CPAN

For alternative/private CPAN, several tools can come to your rescue:

What is not working (2024)

cpanm with cpanfile

Look at following cpanfile:

requires 'Plack', git => 'https://github.com/plack/Plack.git', ref => 'master';

It is not well "honored" by cpanminus.

Broken syntax

requires "MyExperimental::Module" => "git://github.com/lestrrat/p5-MyExperimental-Module.git";

Resources

(some of them are outdated)

[GitHub] Combine Pull Requests... For The Win!

Tib — Thu, 09 Nov 2023 17:03:20 +0000

Picture from Lance Grandahl

In GitHub, whether you want to improve merge velocity, test interdependent changes or setup some sort of releasing model, you could be interested in grouping code changes.

It is called "bundling" or "combining" pull requests.

It allows multiple pull requests to be built and tested together (in CI) and qualified together (e.g. pre-production deployment with shadow traffic).

In order to achieve this easily, you would need some automation. You're lucky, I'm going to share an elegant solution to solve this.

(before we continue, also think about GitHub merge queues)

Demo!

I have 2 pull requests (from forks) targeting main. They have passed minimal tests running on PR but I now would like to run long and consuming tests together. I also want to check they work well together without merging first one of them to main (or to a release branch).

My plan for the two following changes is to target the next release v2.3, so I added a label v2.3 to the pull requests:

In order to trigger the bundle creation, I create the branch v2.3 from main

Once I finished to create branch, it already started to run an action:

After seconds, the run is finished:

A new pull requests (targeting main) appeared:

The pull requests contains well the 2 pull requests changes (and branch v2.3 was updated):

Later, when the bundled pull request will be merged, the combined pull requests will be automatically merged:

Hello "combine-pull-requests"

This magic can happen thanks to the great combine-pull-requests 🎉

Thank you Bob van der Linden for this 👍 🚀

This GitHub action makes it deadly simple to combine multiple pull requests based on their labels.

My preferred workflow consists in triggering bundle by branch creation (but you can use other events) then push and open a pull request to main.

I know you want to see the code, please find below my setup:

name: Combine
on: create

permissions: write-all

jobs:
  combine:
    runs-on: ubuntu-latest
    if: ${{ github.event.ref_type == 'branch'}}
    steps:
      - uses: actions/checkout@v3
        with:
          fetch-depth: 0
      - uses: bobvanderlinden/combine-pull-requests@master
        with:
          label: ${{ github.event.ref }}
          repo-token: ${{ secrets.GITHUB_TOKEN }}
      - name: Push changes
        uses: ad-m/github-push-action@master
        with:
          github_token: ${{ secrets.GITHUB_TOKEN }}
          branch: ${{ github.event.ref }}
      - name: Create pull request
        uses: devops-infra/action-pull-request@v0.4.1
        with:
          github_token: ${{ secrets.GITHUB_TOKEN }}
          title: "[${{ github.event.ref }}] combined!"
          body: ""
          source_branch: ${{ github.event.ref }}
          target_branch: main
          template: .github/BUNDLE_TEMPLATE.md
          label: "bundle

Then you would probably put your build-test-deployment... at the end of this same workflow.

Or you can leave the responsibility to another workflow. There's various manners to differentiate the bundle from other pull requests, for instance you can filter the label bundle.

To give you an example on how you would run/skip a workflow based on PR having the label bundle:

on:
  pull_request:
    types:
      - opened
jobs:
  run_if:
    if: contains(github.event.pull_request.labels.*.name, 'bundle')
    runs-on: ubuntu-latest
    steps:
      - run: echo "This is a bundle, run more tests"

What else?

What if I pushed a new commit to one of the pull request? Just need to delete and recreate branch v2.3. Same procedure applies if I want to add or remove a pull request from the bundle.

Customize to your needs, you can listen to more events than only on: create 😇

Another great thing with labels is that same pull request can belong to multiple bundles 🔥

Conclusion

You should design tooling and branching model that suits you the best, so maybe this is not for you, but I really like how elegant is this workflow and I needed to share.

I hope you can also find it useful for you.

Then, now it only remains for me to wish you "happy hacking!" 😁

Perl Advent Calendar

Tib — Thu, 09 Nov 2023 16:36:36 +0000

Blogging here or there

If in 2023 I did almost no blogging at all, in 2022 I produced a decent number of posts, but not all landed on DevTo!

Late 2022, I helped to make the Perl Advent Calendar tradition to run once again 🎅

2022 be like:

Believe me, all articles in Perl Advent Calendar deserve a read. But let me share mine 😀

This year, I'm not planning to contribute but please give them force and think about contributing yourself!

Open Letter To Tim O'Reilly To Free The Perl Camel

Tib — Fri, 07 Jul 2023 12:11:49 +0000

Dear Tim,

I've been thinking about this letter for a while now.

There was a time (an eternity in IT age) where you did a lot for Perl and I thank you so much for this 🙏

But time passed and Perl has lost its value to your eyes.

Thanks to you ❤️ Perl inherited a logo: the camel
But because of you, Perl actually never had a logo at all 😢

Over the years, you never ever made the Perl camel open enough for us (Perl community) to freely use or modify.
Consequently, we are in a very uncomfortable position.

Let's just use a new logo, what's the point?

No, it's not that simple and we were never able to achieve this.

As of today, we are torn between:

Your camel 🐪 that we are not allowed to use
A shadowed camel 🐪 derived from your camel, still problematic even if it's de facto used everywhere
An onion 🧅 that is not Perl but foundation logo (and not even free to use/modify)
A raptor 🌋 that some people don't like or don't recognize well the association
Various other attempts

It's a long standing problem with numerous discussions, attempts and continuous frustrations for us.

Now, the thing is, only YOU can fix this 🎯

And it's time.

So I'm asking solemnly, free the Perl camel! 🆓
And do it with a nice and clear license.

#FreeThePerlCamel

Please.

Thibault

Notes: it was sent to permissions@oreilly.com

Background

The association of Perl and camel originally comes from Programming Perl book. Since then, O'Reilly always protected it with (too much) restrictions and Perl was never able nor wanted to break association.

We talk about this guy:

...and derivatives!

As far as I know, O'Reilly never enforced restrictions and kept friendly to Perl along all these years.

Past O'Reilly position on Camel

Perl camel usage (webarchive)

Non-commercial use

We will license the camel image widely for open source products and non-commercial sites related to Perl, requiring only an acknowledgement of its trademark status and a link to www.perl.com. To request the camel artwork, please send email to permissions@oreilly.com, indicating where, how, and for what purpose you plan to use the image. Please note that we generally do not allow alterations of the Perl camel artwork.

It sounds free at first but it's not at all 😞

Also note the following sentence (at the beginning) which sounds particularly rude:

After all, the only reason that people think of camels in association with Perl is because we used a camel on the cover of Programming Perl.

Ouch! The last edition of Programming Perl was 11 years ago.

Some clarification in Perl camel FAQ (webarchive)

Q: I want to design a T-shirt with the Perl camel on it. Do I need to get your permission?

A: Yes. But we're willing to make allowances for those of you who have creative ideas and want to do something fun for your friends. So, if the lifetime print run of the T-shirt design is less than 100, you may consider permission automatically granted. For larger print runs, please ask first. We promise to answer quickly!

Q: Why isn't your trademark just restricted to books?

A: We also do conferences, software, research, and online publishing in Perl, and we use the camel image for those things as well. We may want to camel-brand other Perl-related products in the future.

Q: I want to place a picture of a camel on my Perl web page. Am I allowed to do that? Do I have to use your camel?

A: Yes, as long as your page is non-commercial, and the context in which the camel is placed portrays Perl in a positive light. You will need to include the following language in small text somewhere on the page where the camel appears:

"The Perl camel image is a trademark of O'Reilly Media, Inc. Used with permission."

Please make the "O'Reilly Media, Inc." part of the statement a link to our home page (http://www.oreilly.com).

This FAQ sounds terrible to me. It's "yes but no", unclear, restrictive...

At least O'Reilly has some sense of humor:

Q: I want to use $camel as a variable name in a Perl program. Do I need to acknowledge the trademark?

A: No.

Thank you sir, now let's move one step forward and Free The Perl Camel 😉

Recent position

From Perl and Camels, Tim O'Reilly answered questions from Neil Bowers (Perl figure and former perl maintainer). Here's what we can (not) do now:

It's even worse 😠

And what is allowed for shadowed camel?

(It's a derivative of "specific image of the camel that appears on the cover of the Programming Perl book")

Example of discussions

Timeline

06/07/2023 - Sent email "A letter to Tim O'Reilly to ask to free the Perl camel" to permissions@oreilly.com
07/07/2023 - Published blog post
08/07/2023 - Sent another email to permissions@oreilly.com on Saturday
08/07/2023 - I posted on Hacker News and it has hit the front page with 170 votes and 50 comments 🔥 - Open Letter to Tim O’Reilly to Free the Perl Camel
10/07/2023 - Sent email to tim@oreilly.com
14/07/2023 - Got answer by email from O'Reilly

DEV Community: Tib

My participation to the Perl Toolchain Summit 2025

What is the Perl Toolchain Summit?

What about /me?

My plan

Happy Perl Hacking!

Habemus Perl Logo!

The new Perl logo

References

Further on Perl Camel

What's the difference between trademark and copyright

How is trademarked Perl Camel?

Public statement?

USPTO / WIPO

The actual trademark and service mark

Trademark

Service Mark

Trade dress

If copyrights have a duration, how long lasts a trademark?

What is the copyright (license) of O'Reilly Perl camel?

Using Perl shadow camel?

Fair use

Blanket license

Logo vs mascot

My hypothesis and conclusion

My London Perl And Raku Workshop 2024

The review

Visiting around the venue

Fun

The accent problem

UK is same but different

Cultural shift

Errol Court / Error Count

Red Hat

Free library

Perl Monks Pub

Conclusion

App::cpx

Background

App::cpx, a npx-like for CPAN

Your companion for Continuous Integration

Conclusion

Resources

SlapBirdAPM with Mojolicious

Setup Mojolicious Web App

Create SlapBirdAPM account

Install SlapBirdAPM in Web App

Test!

We're done! 😀

3 original Perl CPAN social concepts

CPAN Lieutenants

CPAN Mentors

CPAN Guards

Some reading and watching

The perl.fish experiment

🐟 perl.fish 🐠

A little preview

About the stack

🐟 perl.fish 🐠

Installing CPAN modules from git

Installing from command line with cpm

Installing from command line with cpanminus

Installing from cpanfile

Installing from cpmfile

Beware of "incomplete" repositories

Conclusion

Addendum

Alternative CPAN

What is not working (2024)

cpanm with cpanfile

Broken syntax

Resources

[GitHub] Combine Pull Requests... For The Win!

Demo!

Hello "combine-pull-requests"

What else?

Conclusion

Perl Advent Calendar

Blogging here or there

Open Letter To Tim O'Reilly To Free The Perl Camel

App::cpx, a `npx`-like for CPAN

Installing from command line with `cpm`

Installing from command line with `cpanminus`

Installing from `cpanfile`

Installing from `cpmfile`