How do we push changes into moose - Moose-dev

Tudor Girba

13 Nov 13 Nov

9:27 p.m.

We have absolutely no support for branching other than using a separate Monticello repository and handling the branched code manually. It worked reasonably well until now. I will stress it again. We do need better configuration support, but we need effort spent in this direction, too :). Cheers, Doru On 13 Nov 2011, at 16:23, Stéphane Ducasse wrote:

...

Hi this is still totally unclear to me how I can change Moose and get my changes not breaking parts of moose that I do not know or are not loaded? I want to work on the importer/extractor proposal I sent around last week. So I can implement the changes but what is the process? Stef _______________________________________________ Moose-dev mailing list Moose-dev(a)iam.unibe.ch https://www.iam.unibe.ch/mailman/listinfo/moose-dev

-- www.tudorgirba.com "To utilize feedback, you first have to acquire it."

Reply

Tudor Girba

10:25 p.m.

Hi, On 13 Nov 2011, at 22:10, Stéphane Ducasse wrote:

...

I do not understand the moose process.

What is so difficult to understand? Anytime we can, we load the latest development version (or default how we called it first). If one is not provided, we load some version, like in the case of Merlin.

...

If we cannot control how fixes get introduced in the system and the only solution is to fork then I will fork but this is suboptimal. Ok I will do it then.

I do not understand. Doru

...

Stef

We have absolutely no support for branching other than using a separate Monticello repository and handling the branched code manually. It worked reasonably well until now. I will stress it again. We do need better configuration support, but we need effort spent in this direction, too :). Cheers, Doru On 13 Nov 2011, at 16:23, Stéphane Ducasse wrote:

Hi this is still totally unclear to me how I can change Moose and get my changes not breaking parts of moose that I do not know or are not loaded? I want to work on the importer/extractor proposal I sent around last week. So I can implement the changes but what is the process? Stef _______________________________________________ Moose-dev mailing list Moose-dev(a)iam.unibe.ch https://www.iam.unibe.ch/mailman/listinfo/moose-dev

-- www.tudorgirba.com "To utilize feedback, you first have to acquire it." _______________________________________________ Moose-dev mailing list Moose-dev(a)iam.unibe.ch https://www.iam.unibe.ch/mailman/listinfo/moose-dev

_______________________________________________ Moose-dev mailing list Moose-dev(a)iam.unibe.ch https://www.iam.unibe.ch/mailman/listinfo/moose-dev

-- www.tudorgirba.com "Speaking louder won't make the point worthier."

Reply

Alexandre Bergel

14 Nov 14 Nov

2:46 a.m.

...

What is so difficult to understand? Anytime we can, we load the latest development version (or default how we called it first). If one is not provided, we load some version, like in the case of Merlin.

with merlin this is not the latest since we have to modify the configurationOfMerlin to point to the latest version. But anyway

I suggest to use Versionner for this. Producing a new version of a software with a proper configuration is done with one click. I there is no documentation of Versionner, but it will come soon. A student will start to work on it very soon.

...

If we cannot control how fixes get introduced in the system and the only solution is to fork then I will fork but this is suboptimal. Ok I will do it then.

I do not understand.

since the release cycle of Moose is long and since people use the latest version of Moose it means that this is difficult to introduce large changes. This is probably ok because it forces you to get a working system but if I do a change that touches several packages and one of them does not get loaded or merged then we are in a mess. This is what I mean by how fixes get introduced. In pharo we can **always** rollback in a couple of minutes! Just removing one line in a file and publishing it and we are safe and nobody gets impacted. In Moose no way we can do that. So this is more stress for committers so less people will improve the core.

The problem is that we cannot create a new configuration version of Moose. This is sad, but apparently not easy to fix. Both Doru and I seat down with Dale. Not easy to solve. Alexandre -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.

Reply