Re: [Pharo-project] Pharo 1.3 process
On Apr 3, 2011, at 7:13 PM, laurent laffont wrote:
Hi,
as I'm a little lost on Pharo 1.2 (1.2.1 full was announced but I haven't seen 1.2.0, no big party, no 1 million download contest, BBC announcement... :) I wonder how we can improve the process for 1.3.
Yes, that is why I did not want to decouple the release of Core from Full. Because is drags out the release so far that we can not announce it But we could not, as people (inkluding those maintaining packages of Full) only look at released core images... And so I did a Core 1.2. And people looked at it. Finally. So I did a 1.2.1, and only after releasing that, people looked at it, so I even started a 1.2.2 today.. and now I am dead.
I'm thinking about: - not starting Pharo 1.4 too soon as I feel it will steal energy from Pharo 1.3. - freezing early to have shorter release and have less stuff to fix with maximum energy (it seems energy go down quickly in the fixing period) - a failing test should be highest priority: it's easier to fix a broken feature/test as soon as it appears - one guy should drive the release (I feel that Mariano did this for Pharo 1.1 and we know who we should talk to).
- Get rid of the Core/Full image destinction. This does not work. And I will not do a release with having two images again... it never worked, and it will never work. Marcus -- Marcus Denker -- http://www.marcusdenker.de INRIA Lille -- Nord Europe. Team RMoD.
On Sun, Apr 3, 2011 at 7:18 PM, Marcus Denker <marcus.denker@inria.fr>wrote:
On Apr 3, 2011, at 7:13 PM, laurent laffont wrote:
Hi,
as I'm a little lost on Pharo 1.2 (1.2.1 full was announced but I haven't seen 1.2.0, no big party, no 1 million download contest, BBC announcement... :) I wonder how we can improve the process for 1.3.
Yes, that is why I did not want to decouple the release of Core from Full. Because is drags out the release so far that we can not announce it
But we could not, as people (inkluding those maintaining packages of Full) only look at released core images... And so I did a Core 1.2. And people looked at it. Finally. So I did a 1.2.1, and only after releasing that, people looked at it, so I even started a 1.2.2 today.. and now I am dead.
Time for a big real good fresh beer ;) I'm thinking about:
- not starting Pharo 1.4 too soon as I feel it will steal energy from Pharo 1.3. - freezing early to have shorter release and have less stuff to fix with maximum energy (it seems energy go down quickly in the fixing period) - a failing test should be highest priority: it's easier to fix a broken feature/test as soon as it appears - one guy should drive the release (I feel that Mariano did this for Pharo 1.1 and we know who we should talk to).
- Get rid of the Core/Full image destinction. This does not work. And I will not do a release with having two images again... it never worked, and it will never work.
Yes, 2 images seems hard to maintain and hard to understand for newcomers. May be we can have: - Metacello configurations tested separately in Pharo-Clients in Hudson (like PetitParser, Seaside, ...) - drop the ones that don't have tests - drop actual Pharo, rename PharoCore => Pharo. - GUI tool in Pharo to easy load Metacello configurations (some stuff must be easy to load for newbies, especially help) Laurent. Marcus
-- Marcus Denker -- http://www.marcusdenker.de INRIA Lille -- Nord Europe. Team RMoD.
- Get rid of the Core/Full image destinction. This does not work. And I will not do a release with having two images again... it never worked, and it will never work.
Yes but for that we need a good browser that we can maintain.
Yes, 2 images seems hard to maintain and hard to understand for newcomers.
May be we can have: - Metacello configurations tested separately in Pharo-Clients in Hudson (like PetitParser, Seaside, ...) - drop the ones that don't have tests - drop actual Pharo, rename PharoCore => Pharo. - GUI tool in Pharo to easy load Metacello configurations (some stuff must be easy to load for newbies, especially help
This is the idea. This is why I asked people (mariano in particular) to focus on helping metacello taking off the ground and to focus on metacello distributions. Once this is up and running. we can have Pharo and hudson loads and certifies the distributions. Stef
btw... how's the repository policy we are going to use? to keep MetacelloRepository? or better to use MetacelloRepositoryForPharo12, etc.? if we are going for the 2nd option, we need to start switching :P cheers, Esteban El 03/04/2011, a las 4:17p.m., Stéphane Ducasse escribió:
- Get rid of the Core/Full image destinction. This does not work. And I will not do a release with having two images again... it never worked, and it will never work.
Yes but for that we need a good browser that we can maintain.
Yes, 2 images seems hard to maintain and hard to understand for newcomers.
May be we can have: - Metacello configurations tested separately in Pharo-Clients in Hudson (like PetitParser, Seaside, ...) - drop the ones that don't have tests - drop actual Pharo, rename PharoCore => Pharo. - GUI tool in Pharo to easy load Metacello configurations (some stuff must be easy to load for newbies, especially help
This is the idea. This is why I asked people (mariano in particular) to focus on helping metacello taking off the ground and to focus on metacello distributions. Once this is up and running. we can have Pharo and hudson loads and certifies the distributions.
Stef
second one. Yes I would like but no time. Stef On Apr 4, 2011, at 1:56 PM, Esteban Lorenzano wrote:
btw... how's the repository policy we are going to use? to keep MetacelloRepository? or better to use MetacelloRepositoryForPharo12, etc.?
if we are going for the 2nd option, we need to start switching :P
cheers, Esteban
El 03/04/2011, a las 4:17p.m., Stéphane Ducasse escribió:
- Get rid of the Core/Full image destinction. This does not work. And I will not do a release with having two images again... it never worked, and it will never work.
Yes but for that we need a good browser that we can maintain.
Yes, 2 images seems hard to maintain and hard to understand for newcomers.
May be we can have: - Metacello configurations tested separately in Pharo-Clients in Hudson (like PetitParser, Seaside, ...) - drop the ones that don't have tests - drop actual Pharo, rename PharoCore => Pharo. - GUI tool in Pharo to easy load Metacello configurations (some stuff must be easy to load for newbies, especially help
This is the idea. This is why I asked people (mariano in particular) to focus on helping metacello taking off the ground and to focus on metacello distributions. Once this is up and running. we can have Pharo and hudson loads and certifies the distributions.
Stef
as I'm a little lost on Pharo 1.2 (1.2.1 full was announced but I haven't seen 1.2.0, no big party, no 1 million download contest, BBC announcement... :) I wonder how we can improve the process for 1.3.
Yes, that is why I did not want to decouple the release of Core from Full. Because is drags out the release so far that we can not announce it
But we could not, as people (inkluding those maintaining packages of Full) only look at released core images... And so I did a Core 1.2. And people looked at it. Finally. So I did a 1.2.1, and only after releasing that, people looked at it,
This is not your fault.
so I even started a 1.2.2 today.. and now I am dead.
Let 1.2.2 for in a couple of months.
I'm thinking about: - not starting Pharo 1.4 too soon as I feel it will steal energy from Pharo 1.3. - freezing early to have shorter release and have less stuff to fix with maximum energy (it seems energy go down quickly in the fixing period) - a failing test should be highest priority: it's easier to fix a broken feature/test as soon as it appears - one guy should drive the release (I feel that Mariano did this for Pharo 1.1 and we know who we should talk to).
- Get rid of the Core/Full image destinction. This does not work. And I will not do a release with having two images again... it never worked, and it will never work.
Yes we are working on that. The missing pieces of the puzzle are coming along. Stef
participants (4)
-
Esteban Lorenzano -
laurent laffont -
Marcus Denker -
Stéphane Ducasse