DEV Community: hythm

Pakku Through Images

hythm — Mon, 02 Oct 2023 05:01:16 +0000

Pakku is a package manager for the Raku Programming Language. Latest releases of Pakku are part of Pakku Celastrina version family. Celastrina name means elegant and beutiful, So I will take the opportunity to introduce how elegant IMO Pakku handles Raku distributions.

NOTE
Pakku uses unicode, emojis and colors quite a lot, This might not be everyone's cup of tea, or may be the terminal doesn't support them, so at the end of the post I will show how to disable or customize these features through configurations.

Install Pakku

After clonning Pakku repository, one can run raku -I. bin/pakku add . within the cloned repository to install Pakku:

As shown Pakku installed successfully, and the default verbosity level info was used, which shows the important information for installation steps:

Staging the distribution.
Testing the distribution.
List binaries installed to the system.

If more information during installation is needed, one can raise the verbosity level.

Install Raku Distribution

To install Raku distribution one can run pakku add MyDist

Raising the verbosity to level now makes Pakku output what it is doing at the moment. As shown in the above image, Pakku keeps the user informed when Fetching the distribution, Compiling the distribution modules, also when running the individual tests. and the information flows seamlessly with the progress bar.

More information can be viewed by setting the verbosity level to debug.

There is also all verbosity level, which shows all possible information, including the tests output and much more informations.

here is an example of how it might look when run Pakku with all verbosity level. That might look like a lot of information, but once get used to it, it looks familiar.

Please disregard the dont option in the above image, it basically means do all steps but don't commit any changes to the installation repositories. (equivalent to dry-run in other programs).

Search Raku Distribution

To search Raku distributions one can run pakku search MyDist

The search is relaxed by default, so many dists can be returned (search modules of dist as well), one can pass norelaxed option to narrow the search.

Also if one is interested in latest version only, latest option can be passed to the search command:

Pakku Configurations

The configuration modules consist of:

Pakku general config module.
A module for each Pakku command (eg. add, remove, search etc.).
Log module.
Recommendation manager module.

To view the configs of a module, for instance pakku general configs module:

To disbale colors, bar and spinner, one can run:

pakku config pakku disable pretty bar spinner

Changing colors or replacing emojis with ascii characters can be done via the config command as well, however I will make these changes directly in the config file as below:

And here is how it looks after customization:

To reset configuration:
pakku config reset

To create a new config file:
pakku config new

That's it for now, Thanks for viewing!

Pakku State Update

hythm — Thu, 18 May 2023 18:19:01 +0000

Pakku is a package manager for the Raku Programming Language. In this post I'm going to introduce two Pakku commands, state and update.

state

state command gives an overview of the current state of installed distributions. eg. if there are missing dependencies, if a dist can be updated to a higher version or can be cleaned from the system. etc.

To check the current state of installed Raku dists:

> pakku state
1 - 🦋 STT: ｢...｣
2 - 🦋 UPD: ｢App::Rak:ver<0.2.17>:auth<zef:lizmat>:api<>｣
3 - 🦋 UPD: ｢App::Rak:ver<0.2.16>:auth<zef:lizmat>:api<>｣
4 - 🧚 STT: ｢App::Rak:ver<0.2.15>:auth<zef:lizmat>:api<>｣
...

Reading current state, basically go through all installed dists and check if all deps are met, any updates available, etc.
There is an update available for App::Rak dist.
Same as above but another version.
The current state of installed App::Rak dist is OK, no issues.

🧚 STT: ｢Backtrace::Files:ver<0.0.3>:auth<zef:lizmat>:api<>｣
🧚 STT: ｢CLI::Version:ver<0.0.8>:auth<zef:lizmat>:api<>｣
🧚 STT: ｢Git::Files:ver<0.0.6>:auth<zef:lizmat>:api<>｣
🧚 STT: ｢IO::Path::AutoDecompress:ver<0.0.2>:auth<zef:lizmat>:api<>｣
...

The state of the above dists look fine, no issues, no updates available, no missing deps.

🐞 DEP: ｢JSON::Fast:ver<0.17>:auth<cpan:TIMOTIMO>｣ missing!
🦗 STT: ｢JSON::Fast::Hyper:ver<0.0.3>:auth<zef:lizmat>:api<>｣
...

There is a dependency missing, basically JSON::Fast::Hyper depends on JSON::Fast, but JSON::Fast is not installed. So state of JSON::Fast::Hyper is not OK.

🦋 UPD: ｢String::Utils:ver<0.0.21>:auth<zef:lizmat>:api<>｣
🦋 UPD: ｢String::Utils:ver<0.0.20>:auth<zef:lizmat>:api<>｣
🦋 UPD: ｢String::Utils:ver<0.0.19>:auth<zef:lizmat>:api<>｣
🦋 UPD: ｢String::Utils:ver<0.0.18>:auth<zef:lizmat>:api<>｣
🦋 UPD: ｢String::Utils:ver<0.0.17>:auth<zef:lizmat>:api<>｣
🦋 UPD: ｢String::Utils:ver<0.0.16>:auth<zef:lizmat>:api<>｣
🧚 STT: ｢String::Utils:ver<0.0.15>:auth<zef:lizmat>:api<>｣
...

There are multiple updates available for installed String::Utils dist.

1 - 🦋 UPD: ｢paths:ver<10.0.9>:auth<zef:lizmat>:api<>｣
2 - 🧚 STT: ｢paths:ver<10.0.8>:auth<zef:lizmat>:api<>｣
3 - 🦋 CLN: ｢paths:ver<10.0.8>:auth<zef:lizmat>:api<>｣
4 - 🧚 STT: ｢paths:ver<10.0.9>:auth<zef:lizmat>:api<>｣
5 - 🧚 STT: ｢rak:ver<0.0.47>:auth<zef:lizmat>:api<>｣
...

There is an update available for paths dist
the current state of paths:ver<10.0.8> dist is OK
paths:ver<10.0.8> dist can be cleand, this means there is a higher version already installed (which is evident by next line), and no other dists depend on it.
paths:ver<10.0.9> dist state is OK
rak dist state is OK

Now I have an overview of the current state of the system, and if more info needed, I can run in debug mode and see more information like, dist dependencies, reverse dependencies, etc.

With this info I can install the missing deps or clean not needed dists. I also can update the system to get latest available versions, which what I'll do in the next section of this post.

update

update command is used for updating all installed dists or specific dist to latest available version.

From previous section I found out that there are updates for App::Rak and String::Utils dists, there is also a missing dependency JSON::Fast:ver<0.17>:auth<cpan:TIMOTIMO> and some dists can be cleaned.

So time to run the update command:

> pakku update
...
🦋 STT: ｢...｣
🦋 UPD: ｢App::Rak:ver<0.2.17>:auth<zef:lizmat>:api<>｣
🦋 UPD: ｢String::Utils:ver<0.0.21>:auth<zef:lizmat>:api<>｣
🦋 UPD: ｢paths:ver<10.0.9>:auth<zef:lizmat>:api<>｣
...

The update command starts by reading the current state, identify the updatable dists and resolve deps...

🧚 TST: ｢JSON::Fast:ver<0.17>:auth<cpan:TIMOTIMO>:api<>｣
🦋 STG: ｢App::Rak:ver<0.2.17>:auth<zef:lizmat>:api<>｣
🦋 TST: ｢App::Rak:ver<0.2.17>:auth<zef:lizmat>:api<>｣
🦋 TST: ｢01-basic.rakutest｣
🧚 TST: ｢App::Rak:ver<0.2.17>:auth<zef:lizmat>:api<>｣
🦋 STG: ｢String::Utils:ver<0.0.21>:auth<zef:lizmat>:api<>｣
🦋 TST: ｢String::Utils:ver<0.0.21>:auth<zef:lizmat>:api<>｣
🦋 TST: ｢01-basic.rakutest｣
🦋 TST: ｢02-selective-importing.t｣
🧚 TST: ｢String::Utils:ver<0.0.21>:auth<zef:lizmat>:api<>｣
🧚 BIN: ｢rak｣

Installed JSON::Fast missing dep, updated App::Rak and String::Utils

🦋 RMV: ｢App::Rak:ver<0.2.15>:auth<zef:lizmat>:api<> paths:ver<10.0.8>:auth<zef:lizmat>:api<>｣
🦋 RMV: ｢App::Rak:ver<0.2.15>:auth<zef:lizmat>:api<>｣
🦋 RMV: ｢paths:ver<10.0.8>:auth<zef:lizmat>:api<>｣
🧚 -Ofun

Cleaning the not needed dists, the older versions of App::Rak and paths.
One might notice that String::Utils:ver<0.0.15> was not cleaned, and that is because App::Rak depends on that specific version.

If one doesn't want to clean older versions of dists, noclean option can be provided to update command.

Thanks for reading!

Unpacking Pakku

hythm — Sun, 07 May 2023 17:43:55 +0000

In this post I'm going to explain how I use Pakku (a package manager for Raku) to manage Raku distributions. On a clean Rakudo installation, I will start with installing Pakku, then use some of the features like installing a distribution, search the ecosystem for a module and other features like configuring a local recommendation manager. and in the process I will explain how Pakku works.

Bootstrapping is easy

The first thing I do on a new Rakudo installation is I install Pakku. all I need to do is clone the repo and ask Pakku to install Pakku!

> git clone https://github.com/hythm7/Pakku.git
> cd Pakku
> raku -I. bin/pakku add .
🧚 PRC: ｢.｣
🦋 STG: ｢Pakku:ver<caria>:auth<github:hythm7>:api<0>｣
🦋 TST: ｢Pakku:ver<caria>:auth<github:hythm7>:api<0>｣
🦋 TST: ｢cache.rakutest｣
🦋 TST: ｢cmd.rakutest｣
🦋 TST: ｢config.rakutest｣
🦋 TST: ｢core.rakutest｣
🦋 TST: ｢log.rakutest｣
🦋 TST: ｢meta.rakutest｣
🦋 TST: ｢spec.rakutest｣
🧚 TST: ｢｢Pakku:ver<caria>:auth<github:hythm7>:api<0>｣
🧚 BIN: ｢pakku｣
🧚 -Ofun

The output shows that Pakku distribution was staged, then test files were run, and also a binary pakku was added. We will go through these different stages in "Installing a distribution" section later in this post.

But for now looks like Pakku can install it self, thats nice. however, if one looked at the installation code will realize that the actual installation logic lives in Rakudo here, and that what made bootstrapping Pakku easy.

Searching for a module

To search the ecosystems for a module, for instance a DB module:

> pakku search db
｢DB:ver<0.5>:auth<github:CurtTilmes>:api<1>｣
｢DB:ver<0.4>:auth<>:api<>｣
｢DB:ver<0.3>:auth<>:api<>｣
｢DB:ver<0.2>:auth<>:api<>｣
｢DB:ver<0.1>:auth<>:api<>｣
｢DB::Migration::Declare:ver<0.1.3>:auth<zef:jnthn>:api<>｣
｢DB::Migration::Declare:ver<0.1.2>:auth<zef:jnthn>:api<>｣
｢DB::Migration::Declare:ver<0.1.1>:auth<zef:jnthn>:api<>｣
｢DB::Migration::Declare:ver<0.1>:auth<zef:jnthn>:api<>｣
｢DB::Migration::Simple:ver<1.0>:auth<>:api<>｣
｢DB::Model::Easy:ver<*>:auth<>:api<>｣
｢DB::MySQL:ver<0.5>:auth<>:api<>｣
...

The search is relaxed by default, but if exact match is needed, one can provide norelaxed option:

> pakku search norelaxed db
｢DB:ver<0.5>:auth<github:CurtTilmes>:api<1>｣
｢DB:ver<0.4>:auth<>:api<>｣
｢DB:ver<0.3>:auth<>:api<>｣
｢DB:ver<0.2>:auth<>:api<>｣
｢DB:ver<0.1>:auth<>:api<>｣
🧚 -Ofun

One can check the details of a distribution, such as description, dependencies and what it provides. for example:

> pakku search details 'DB::Migration::Declare:ver<0.1.3>'
MTA: ｢DB::Migration::Declare:ver<0.1.3>:auth<zef:jnthn>:api<>｣
DEP: ｢Digest::SHA1::Native｣
DEP: ｢Test::ContainerizedService｣
PRV: ｢DB::Migration::Declare｣
PRV: ｢DB::Migration::Declare::Applicator｣
PRV: ｢DB::Migration::Declare::ColumnType｣
PRV: ｢DB::Migration::Declare::Database｣
PRV: ｢DB::Migration::Declare::Database::Postgres｣
PRV: ｢DB::Migration::Declare::MigrationDirection｣
PRV: ｢DB::Migration::Declare::MigrationHistory｣
PRV: ｢DB::Migration::Declare::Model｣
PRV: ｢DB::Migration::Declare::Model::MigrationStep｣
PRV: ｢DB::Migration::Declare::Problem｣
PRV: ｢DB::Migration::Declare::SQLLiteral｣
PRV: ｢DB::Migration::Declare::Schema｣
PRV: ｢DB::Migration::Declare::Test｣
URL: ｢https://github.com/jnthn/db-migration-declare.git｣
DES: ｢Specify database migrations using a Raku DSL. Currently Postgres only.｣
🧚 -Ofun

Installing a distribution

In this section I will install Inline::Perl5 in debug mode enabled. because Inline::Perl5 has multiple dependencies (including bin dependency), and a build stage, I think it's a good candidate for this section.

since running in debug mode, I will truncate the output as possible for easy readability, then explain the steps Pakku goes through for installing a distribution.

> pakku verbose debug add Inline::Perl5
1 - 🧚 PRC: ｢Inline::Perl5｣
2 - 🐛 SPC: ｢Inline::Perl5｣
3 - 🐛 CAC: ｢Inline::Perl5｣ recommending...
4 - 🐛 REC: ｢pakku｣ ‹Inline::Perl5› recommending...
5 - 🦋 MTA: ｢Inline::Perl5:ver<0.60>:auth<github:niner>:api<>｣

1 - Started processing Inline::Perl5.

2 - Recieved Inline::Perl5 specification.

3 - Asking the cache for a recommendation for Inline::Perl5 spec.

4 - No recommendation from cache, so asking the default recommendation manager (named pakku).

5 - Inline::Perl5:ver<0.60>:auth<github:niner>:api<> meta info recommended from the recommendation manager.

🐛 SPC: ｢Distribution::Builder::MakeFromJSON:ver<0.6+>｣
🐛 CAC: ｢Distribution::Builder::MakeFromJSON:ver<0.6+>｣ recommending...
🐛 REC: ｢pakku｣ ‹Distribution::Builder::MakeFromJSON:ver<0.6+>› recommending...
🦋 MTA: ｢Distribution::Builder::MakeFromJSON:ver<0.6>:auth<github:niner>:api<>｣
...

The meta info is checked for dependencies, and run the same steps for found dependencies.

🐛 SPC: ｢perl:from<bin>｣ satisfied!
🐛 SPC: ｢JSON::Fast｣ satisfied!
🐛 SPC: ｢Test｣ satisfied!
...

Nothing will be done for any of these dependencies specs because they are already satisfied. Confirmed that perl binary exists in the system. JSON::Fast is already installed. Test is already installed from rakudo core distribution.

🦋 FTC: ｢Distribution::Builder::MakeFromJSON:ver<0.6>:auth<github:niner>:api<>｣
🐛 CAC: ｢Distribution::Builder::MakeFromJSON:ver<0.6>:auth<github:niner>:api<>｣ looking...
🐛 FTC: ｢http://truncated...｣
🐛 FTC: ｢/home/hythm/.pakku/.tmp/truncated...｣
🐛 CAC: ｢Distribution::Builder::MakeFromJSON:ver<0.6>:auth<github:niner>:api<>｣ caching...
🐛 CAC: ｢Distribution::Builder::MakeFromJSON:ver<0.6>:auth<github:niner>:api<>｣ ‹/home/hythm/.pakku/.cache/truncated...›
...

Now as all dependencies have been resolved, it is time to fetch the distributions. so first, check if the distribution is in the cache, if not then fetch the archive, extract it to a temp directory, and cache it so it can be found in cache next time.

...
1 - 🦋 BLD: ｢Inline::Perl5:ver<0.60>:auth<github:niner>:api<>｣
2 - 🐛 BLD: ｢/home/hythm/.rakubrew/versions/moar-blead/install/bin/rakudo truncated...
3 - 🐛 BLD: gcc -Wall p5helper.c -Wl,-E  -fstack-protector-strong -L/usr/ truncated...
4 - 🧚 BLD: ｢Inline::Perl5:ver<0.60>:auth<github:niner>:api<>｣
5 - 🦋 STG: ｢Inline::Perl5:ver<0.60>:auth<github:niner>:api<>｣
6 - 🦋 TST: ｢Inline::Perl5:ver<0.60>:auth<github:niner>:api<>｣
7 - ...
8 - 🧚 TST: ｢Inline::Perl5:ver<0.60>:auth<github:niner>:api<>｣
🧚 -Ofun

Now as all distributions have been fetched, what is remaining is to build, stage, test and install them.

1 - Start building

2 - The running build command

3 - The build process output

4 - Build succeeded

5 - Staging the distribution - installing the dist to a staging repository

6 - Start testing the distribution using the staging repository.

7 - The truncated output shows the files being tested and their output

8 - Test succeeded

After all dists have been tested successfully, the staging repository content is moved to the actuall installation repo finishing the installation process.

List installed dists

To list all installed dists, one can run:

> pakku list
｢Libarchive:ver<0.1>:auth<github:CurtTilmes>:api<>｣
｢Number::Bytes::Human:ver<0.0.3>:auth<>:api<>｣
｢JSON::OptIn:ver<0.0.2>:auth<zef:jonathanstowe>:api<>｣
｢Cro::Core:ver<0.8.9>:auth<zef:cro>:api<>｣
...

To list installed dists in a specific repo, eg. core:

> pakku list repo core
｢rakudo:ver<2023.04.45.g.5965.d.7.fb.0>:auth<Yet Another Society>:api<>｣
🧚 -Ofun

To check if a dist is installed:

> pakku list JSON::Fast
｢JSON::Fast:ver<0.17>:auth<cpan:TIMOTIMO>:api<>｣
🧚 -Ofun

Recommendation managers

Pakku currently supports two types of recommendation managers:

1- HTTP service recommendation manager

Pakku ships configured with a default HTTP recman, that has all available Raku distriutions in Raku Ecosystem Archive

Can check the configured recommendation managers by running:

> pakku config recman
🦋 CNF: ｢recman｣
[
  {
    "active": true,
    "location": "http://recman.pakku.org",
    "name": "pakku",
    "priority": 1
  }
]
🧚 -Ofun

2- Local recommendation manager

Used for making a recommendation manager from a local directory containing a collection of Raku dists, and these dists can be seen by Pakku and become available for installation.

To configure a local recman one can run:

> pakku config recman name-of-local-recman set location /home/hythm/dev/my-collection-of-raku-dists
🦋 CNF: ｢recman｣
🦋 REC: ｢name-of-local-recman｣
🦋 CNF: ｢{"location":"/home/hythm/dev/my-collection-of-raku-dists"}｣
🧚 -Ofun

A local recman is a quick and easy way to turn a local collection of dists into a recman, but if one needs more managable way (eg manage dists internally in a company or an organization) it's advised to use an HTTP service recamn. in this case the HTTP service recman expects a collection of dist archives, and it will process them, build a database and can serve and recommend the dists. this can be acheived using recman-simple distribution.

When multiple recmans are configured, it's advised to set the priority of each one as needed, because Pakku searches recmans based on priority, if nothing recommended from first recman, then search the next one.

To set recman priority:

> pakku config recman name-of-local-recman set priority 2
🦋 CNF: ｢recman｣
🦋 REC: ｢name-of-local-recman｣
🦋 CNF: ｢{"priority":"2"}｣
🧚 -Ofun