Released versions in Pharo Bootstrap

Hi,

it is an issue of Calypso dependencies. I discussed it with Denis
before Calypso integration and he would like to change it but, for
now, it would make the development for him much harder because he
needs a way how to load the latest bleeding edge versions of the
packages and no one was able to tell him the proper way how to support
both scenarios (loading of a specific release and loading of the
development versions) with baselines without needing of duplicate
them.

Anyway, I agree that the current state must be fixed as soon as possible.

Cheers,
-- Pavel

2018-03-05 15:30 GMT+01:00 [hidden email] <[hidden email]>:
> Hello,
>      i have seen in the latest version of Pharo baselines pointing to
> "floating" versions. A version that is not fixed. I want to know why this is
> like that?
>
> To me this is basically wrong, because basically this goes against the idea
> of having repeatable builds. Also using semantic versions with tags in Git
> is wrong, as tags should be immutable (that is why you have to use --force
> to update them, in Guille's words "If you have to force it, it is not
> good").
>
> Cheers,
>
> --
> Pablo Tesone.
> [hidden email]

Yes, I understand for the developer version, you should have a developer branch and use that. The problem is for the Pharo image only. 

On Mon, Mar 5, 2018 at 3:39 PM, Pavel Krivanek <[hidden email]> wrote:

Hi,

it is an issue of Calypso dependencies. I discussed it with Denis
before Calypso integration and he would like to change it but, for
now, it would make the development for him much harder because he
needs a way how to load the latest bleeding edge versions of the
packages and no one was able to tell him the proper way how to support
both scenarios (loading of a specific release and loading of the
development versions) with baselines without needing of duplicate
them.

Anyway, I agree that the current state must be fixed as soon as possible.

Cheers,
-- Pavel

2018-03-05 15:30 GMT+01:00 [hidden email] <[hidden email]>:
> Hello,
>      i have seen in the latest version of Pharo baselines pointing to
> "floating" versions. A version that is not fixed. I want to know why this is
> like that?
>
> To me this is basically wrong, because basically this goes against the idea
> of having repeatable builds. Also using semantic versions with tags in Git
> is wrong, as tags should be immutable (that is why you have to use --force
> to update them, in Guille's words "If you have to force it, it is not
> good").
>
> Cheers,
>
> --
> Pablo Tesone.
> [hidden email]

--

Pablo Tesone.
[hidden email]

Yes, I understand for the developer version, you should have a developer branch and use that. The problem is for the Pharo image only. 

On Mon, Mar 5, 2018 at 3:39 PM, Pavel Krivanek <[hidden email]> wrote:

Hi,

it is an issue of Calypso dependencies. I discussed it with Denis
before Calypso integration and he would like to change it but, for
now, it would make the development for him much harder because he
needs a way how to load the latest bleeding edge versions of the
packages and no one was able to tell him the proper way how to support
both scenarios (loading of a specific release and loading of the
development versions) with baselines without needing of duplicate
them.

Anyway, I agree that the current state must be fixed as soon as possible.

Cheers,
-- Pavel

2018-03-05 15:30 GMT+01:00 [hidden email] <[hidden email]>:
> Hello,
>      i have seen in the latest version of Pharo baselines pointing to
> "floating" versions. A version that is not fixed. I want to know why this is
> like that?
>
> To me this is basically wrong, because basically this goes against the idea
> of having repeatable builds. Also using semantic versions with tags in Git
> is wrong, as tags should be immutable (that is why you have to use --force
> to update them, in Guille's words "If you have to force it, it is not
> good").
>
> Cheers,
>
> --
> Pablo Tesone.
> [hidden email]

--

Pablo Tesone.
[hidden email]

2018-03-05 16:13 GMT+01:00 Cyril Ferlicot <[hidden email]>:

On Mon, Mar 5, 2018 at 4:10 PM, Denis Kudriashov <[hidden email]> wrote:
> Hi Pablo.
>
> Dev branch approach not really works. Because any merge into master will
> break master baseline. (notice that baseline is in same repo).
> And managing merges by hand all the time is not a solution.
>
>
Hi!
If you don't want to manage the merges by hand you can maybe have two bsaelines?
BaselineOfCalypso and BaselineOfCalypsoDev?

It should work. But is it right way that everybody should follow?

With configurations it was easy to do in single class.  

 

--
Cyril Ferlicot
https://ferlicot.fr

But, "one single class" does not mean anything. Because it depends from which branch/commitish you are loading it from...

Can you explain better what is the problem because I am not getting it...

In any case, independently of where is the burden, I want to veto any new integration that may make future builds non-reproducible. Otherwise this is a source of chaos and dead kittens.

On Mon, Mar 5, 2018 at 4:17 PM, Denis Kudriashov <[hidden email]> wrote:

2018-03-05 16:13 GMT+01:00 Cyril Ferlicot <[hidden email]>:

On Mon, Mar 5, 2018 at 4:10 PM, Denis Kudriashov <[hidden email]> wrote:
> Hi Pablo.
>
> Dev branch approach not really works. Because any merge into master will
> break master baseline. (notice that baseline is in same repo).
> And managing merges by hand all the time is not a solution.
>
>
Hi!
If you don't want to manage the merges by hand you can maybe have two bsaelines?
BaselineOfCalypso and BaselineOfCalypsoDev?

It should work. But is it right way that everybody should follow?

With configurations it was easy to do in single class.  

 

--
Cyril Ferlicot
https://ferlicot.fr

--

Guille Polito Research Engineer Centre de Recherche en Informatique, Signal et Automatique de Lille CRIStAL - UMR 9189 French National Center for Scientific Research - http://www.cnrs.fr Web: http://guillep.github.io Phone: <a href="tel:+33%206%2052%2070%2066%2013" value="+33652706613" target="_blank">+33 06 52 70 66 13

Maybe we should just load Calypso dependencies explicitly.

In any case they will be used by other projects. Like new Iceberg will use Commander. And ClassAnnotation will be in Kernel.

So we will need to load them in another time than Calypso.

Now we can just put them directly into #loadCalypso method. (there are three dependencies)

It will fix my problem and make Pharo build reproducible.

2018-03-05 16:38 GMT+01:00 Guillermo Polito <[hidden email]>:

But, "one single class" does not mean anything. Because it depends from which branch/commitish you are loading it from...

Can you explain better what is the problem because I am not getting it...

In any case, independently of where is the burden, I want to veto any new integration that may make future builds non-reproducible. Otherwise this is a source of chaos and dead kittens.

On Mon, Mar 5, 2018 at 4:17 PM, Denis Kudriashov <[hidden email]> wrote:

2018-03-05 16:13 GMT+01:00 Cyril Ferlicot <[hidden email]>:

On Mon, Mar 5, 2018 at 4:10 PM, Denis Kudriashov <[hidden email]> wrote:
> Hi Pablo.
>
> Dev branch approach not really works. Because any merge into master will
> break master baseline. (notice that baseline is in same repo).
> And managing merges by hand all the time is not a solution.
>
>
Hi!
If you don't want to manage the merges by hand you can maybe have two bsaelines?
BaselineOfCalypso and BaselineOfCalypsoDev?

It should work. But is it right way that everybody should follow?

With configurations it was easy to do in single class.  

 

--
Cyril Ferlicot
https://ferlicot.fr

--

Guille Polito Research Engineer Centre de Recherche en Informatique, Signal et Automatique de Lille CRIStAL - UMR 9189 French National Center for Scientific Research - http://www.cnrs.fr Web: http://guillep.github.io Phone: <a href="tel:+33%206%2052%2070%2066%2013" value="+33652706613" target="_blank">+33 06 52 70 66 13

But can you explain me the scenario? I still don't understand it :(

On Mon, Mar 5, 2018 at 4:51 PM, Denis Kudriashov <[hidden email]> wrote:

Maybe we should just load Calypso dependencies explicitly.

In any case they will be used by other projects. Like new Iceberg will use Commander. And ClassAnnotation will be in Kernel.

So we will need to load them in another time than Calypso.

Now we can just put them directly into #loadCalypso method. (there are three dependencies)

It will fix my problem and make Pharo build reproducible.

2018-03-05 16:38 GMT+01:00 Guillermo Polito <[hidden email]>:

But, "one single class" does not mean anything. Because it depends from which branch/commitish you are loading it from...

Can you explain better what is the problem because I am not getting it...

In any case, independently of where is the burden, I want to veto any new integration that may make future builds non-reproducible. Otherwise this is a source of chaos and dead kittens.

On Mon, Mar 5, 2018 at 4:17 PM, Denis Kudriashov <[hidden email]> wrote:

2018-03-05 16:13 GMT+01:00 Cyril Ferlicot <[hidden email]>:

On Mon, Mar 5, 2018 at 4:10 PM, Denis Kudriashov <[hidden email]> wrote:
> Hi Pablo.
>
> Dev branch approach not really works. Because any merge into master will
> break master baseline. (notice that baseline is in same repo).
> And managing merges by hand all the time is not a solution.
>
>
Hi!
If you don't want to manage the merges by hand you can maybe have two bsaelines?
BaselineOfCalypso and BaselineOfCalypsoDev?

It should work. But is it right way that everybody should follow?

With configurations it was easy to do in single class.  

 

--
Cyril Ferlicot
https://ferlicot.fr

--

Guille Polito Research Engineer Centre de Recherche en Informatique, Signal et Automatique de Lille CRIStAL - UMR 9189 French National Center for Scientific Research - http://www.cnrs.fr Web: http://guillep.github.io Phone: <a href="tel:+33%206%2052%2070%2066%2013" value="+33652706613" target="_blank">+33 06 52 70 66 13

--

Guille Polito Research Engineer Centre de Recherche en Informatique, Signal et Automatique de Lille CRIStAL - UMR 9189 French National Center for Scientific Research - http://www.cnrs.fr Web: http://guillep.github.io Phone: <a href="tel:+33%206%2052%2070%2066%2013" value="+33652706613" target="_blank">+33 06 52 70 66 13

2018-03-05 17:02 GMT+01:00 Cyril Ferlicot <[hidden email]>:

On Mon, Mar 5, 2018 at 4:51 PM, Guillermo Polito <[hidden email]> wrote:

I still do not understand... Also I do not understand your usage of the term "the merge". Can somebody give a **concrete** scenario? 

I'll try

For example the project MaterialDesignLite have two branches that are always here:

- master 

- development 

Each commit on master is a stable release and end up with a tag as "v1.2.2".

I do not agree with this sentence: 

 

Since it's stable, BaselineOfMaterialDesignLite should depend only on fixed version of the dependencies. For example it should depend on MaterialDesignColor "v1.0.0".

Maybe I am wrong of course. But this is what I expect from semantic versioning:

If I made project which uses some libraries I want to get latest fixes even in my stable version. This is what semantics versioning is about. Minor releases are supposed to be safe for dependent projects.

It is of course question of trust to third parties. But if I do not trust them I should fork dependency and use it instead of original one.

On the branch dev, we want to get the patches and possibly the minor versions of the dependencies automatically. In the baseline we then want to depend on MaterialDesignColor "v1.x.x". Thus, we follow the changes of MaterialDesignColor while it's not a major release.

Because of this situation, BaselineOfMaterialDesignLite is different on the two branches. Later, if I want to merge development into master in order to release a new version, master will get the BaselineOfMaterialDesignLite with semantic versionning dependencies instead of the fixed dependencies. Before the release I'll need to change the Baseline to get fix dependencies once again. 

I hope I was clearer. :)

 

When you release a version, please do not move that version. You should then create new versions, with new numbers and new code. But never touch old versions with old numbers and old code. Like that I can download the same old code using the same old number to get the old version whenever I want :).

You can try to do it with branches, tags, different repositories, or even with zipfiles in mails. I don't care. Just don't modify releases and I'm happy with it.

 

--

Cyril Ferlicot
https://ferlicot.fr

The well known problem with fixed configurations is dependencies:

My project A version 1.2.3 depends on project C version 4.3 (semantic versioning). I have tested it with 4.3.35, it works well...
If semantic versioning is correctly used, it should work with any 4.x.y where x>=3.

There is another project B version 2.1.4 which depends on project C version 4.4 (still semantic).

Loading latest patch 4.4.20 should work for both A & B.
And even 4.6.11 which is the latest compatible version in 4.x serie

IMO, the package ConfigurationOfPackageA and ConfigurationOfPackageB should NOT specify the exact version of package C, but only the minimal compatible version.

Then, if you want a reproducible image, that is up to the specific assembly of package A and package B (let's call it ConfigurationOfPackageAandBandC) that you should force the specific C version.

Isn't it the problem?

2018-03-05 17:17 GMT+01:00 Denis Kudriashov <[hidden email]>:

2018-03-05 17:02 GMT+01:00 Cyril Ferlicot <[hidden email]>:

On Mon, Mar 5, 2018 at 4:51 PM, Guillermo Polito <[hidden email]> wrote:

I still do not understand... Also I do not understand your usage of the term "the merge". Can somebody give a **concrete** scenario? 

I'll try

For example the project MaterialDesignLite have two branches that are always here:

- master 

- development 

Each commit on master is a stable release and end up with a tag as "v1.2.2".

I do not agree with this sentence: 

 

Since it's stable, BaselineOfMaterialDesignLite should depend only on fixed version of the dependencies. For example it should depend on MaterialDesignColor "v1.0.0".

Maybe I am wrong of course. But this is what I expect from semantic versioning:

If I made project which uses some libraries I want to get latest fixes even in my stable version. This is what semantics versioning is about. Minor releases are supposed to be safe for dependent projects.

It is of course question of trust to third parties. But if I do not trust them I should fork dependency and use it instead of original one.

On the branch dev, we want to get the patches and possibly the minor versions of the dependencies automatically. In the baseline we then want to depend on MaterialDesignColor "v1.x.x". Thus, we follow the changes of MaterialDesignColor while it's not a major release.

Because of this situation, BaselineOfMaterialDesignLite is different on the two branches. Later, if I want to merge development into master in order to release a new version, master will get the BaselineOfMaterialDesignLite with semantic versionning dependencies instead of the fixed dependencies. Before the release I'll need to change the Baseline to get fix dependencies once again. 

I hope I was clearer. :)

 

When you release a version, please do not move that version. You should then create new versions, with new numbers and new code. But never touch old versions with old numbers and old code. Like that I can download the same old code using the same old number to get the old version whenever I want :).

You can try to do it with branches, tags, different repositories, or even with zipfiles in mails. I don't care. Just don't modify releases and I'm happy with it.

 

--

Cyril Ferlicot
https://ferlicot.fr