Amber packaging and loading (was: Re: [amber-lang] How to organize Amber extensions/addons)

> Hi all,
>
> I started Amber last autumn and I started to share some of the basic
> stuff I successfully extended Amber with on Github.
>
> I'll soon have some more things to share and I was wondering which way
> might be best.
>
> Could we setup "one" general kind of "solution" to provide Amber
> extensions?
>
> I'm not too sure if I should use, Github, Git submodules, an examples
> folder subfolder on github, Smalltalkhub with Kaliningrad, ss3 with
> Kaliningrad,... pull requests to the general Amber repository to add
> stuff to the examples folder directly,...
> Will there soon be a solution which might be included in Pharo?...
>
> What are your preferences? How do you share? Who's to ask and who's to
> blame? ;-)
>
> Cheers!
> Sebastian

> Hi,
>
> very good topic.
>
> I myself struggle with this, even in a simple scenario. When writing
> library for Amber, I'd like to have Amber itself, the library and also
> test/example/demo code available.
>
> For now, I could have two pretty easily (library plus Amber itself), but
> it is hard to have more. Not to mention the possibility that my library
> would depend on another library.
>
> So I think the loader should be updated, to somehow allow loading
> packages from more locations (and maybe store them as well, though
> generally it would be sufficient if only one of the tier would be
> writable, but we need the ability to have more of them for reading).
>
> I personally think the loader should allow to put Amber libraries into
> npm (thus, dependencies solved on server / deploy side), and have the
> possinility to load them from the npm structure.
>
> A fairly easy, general solution could be instead:
>
>    loadAmber({
>      packages: ['Foo', 'Bar'],
>      packageHome: '../lib'
>    });
>
> to be able to use:
>
>    loadAmber({
>      packages: {
>        '../lib': ['Foo', 'Bar', { './relative': 'Baz' }],
>        '/another/path': [ 'Quux' ]
>      }
>    });
>
> IOW, to put free structure in packages and its subelements where:
>    - if it is a string, it is the package name itself ('Baz')
>    - if it is an array, it is list of elements
>    - if it is an object (aka hash), it maps paths, relative to actual
> path, to subelements.
>
> But anything other would do as well.
>
> Herby
>
> P.S.: Sorry for occassional b changed to n, I've got fresh natural
> keyboard and must relearn that b should be typed by left hand. ;-)
>
> Sebastian Heidbrink wrote:
>> Hi all,
>>
>> I started Amber last autumn and I started to share some of the basic
>> stuff I successfully extended Amber with on Github.
>>
>> I'll soon have some more things to share and I was wondering which way
>> might be best.
>>
>> Could we setup "one" general kind of "solution" to provide Amber
>> extensions?
>>
>> I'm not too sure if I should use, Github, Git submodules, an examples
>> folder subfolder on github, Smalltalkhub with Kaliningrad, ss3 with
>> Kaliningrad,... pull requests to the general Amber repository to add
>> stuff to the examples folder directly,...
>> Will there soon be a solution which might be included in Pharo?...
>>
>> What are your preferences? How do you share? Who's to ask and who's to
>> blame? ;-)
>>
>> Cheers!
>> Sebastian

> I did.
>
> Nicolas Petton wrote:
>> A good start would be to write ideas on a package loader on the wiki of
>> Amber on github.
>>
>> Herby, do you want to start it? I like your ideas :)
>>
>> Nico
>