maven-doxia-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Lukas Theussl <ltheu...@apache.org>
Subject Re: What should go into doxia 1.0-alpha-10?
Date Wed, 31 Oct 2007 13:02:45 GMT
[somehow I managed to misplace your mail, sorry for the delay]

Carlos Sanchez wrote:
> I like your comments but let me argue a bit ;)

arguing is good! :)

> 
> - what is what makes something "stable" ? dramatic API changes can go
> in 2.0, 3.0,... they don't need to go before 1.0 (if not there would
> never be any 1.0 final in any project)

Major API changes IMO can go anywhere except point/bug-fix versions of 
stable releases. I agree that we don't *have* to put everything into 
1.0. However, fact is that current doxia is really just useful for doing 
one thing: translate apt and fml into xhtml, which is the maven use 
case. Try to do anything else and you're sure to run into trouble. Just 
look at the issues people have eg with twiki and confluence modules.

My initial motivation for getting involved with doxia was to write a pdf 
plugin for m2. The current sandbox version is basically ready by itself, 
but it's unusable in practice because of serveral shortcomings of doxia 
in other places (DOXIA-148, DOXIA-183). Doxia promises to be a 
multi-format content generation framework, as long as it doesn't live up 
to this promise, it doesn't deserve a 1.0 release IMO.

> - if something has been out for months/years and did not change
> drastically, maybe should be considered stable

So all you have to do to make a crappy product stable is sit back and 
wait long enough? ;)

> 
> - being used Maven "stable" releases for 2+ years doesn't mean it's stable?
> or
> - if it's not stable maybe we shouldn't be using it in stable versions of Maven?

Personally, I would go with the latter ;) But as I said, Doxia is all 
geared to work with Maven (the site-generating plugins), if that's all 
it is meant to do, then you can consider it stable. But I don't think 
that's all it is about.

> 
> I just think there's a history of aversion to final releases that we
> should get over

I agree. There has to be a point where you need to let your baby go, but 
not before it's able to support itself... My expectation is that we have 
at most one beta and maybe some rc's before 1.0-final.

Cheers,
-Lukas

> 
> 
> On 10/25/07, Lukas Theussl <ltheussl@apache.org> wrote:
> 
>>Carlos Sanchez wrote:
>>
>>>you guys realize that final versions of Maven have been using Doxia
>>>for quite some time now and the fact that it's being called alpha as
>>>an excuse to make more API changes, which will affect the future
>>>development of maven, instead of evolving the API in a backwards
>>>compatible way, it's not a good idea at all
>>>
>>>my 0.02
>>>
>>
>>Doxia is not called alpha as an excuse for more API changes, it's called
>>alpha because it *is* alpha. Just have a look at the code. Have a look
>>at some open issues [eg DOXIA-38, DOXIA-63, DOXIA-78, DOXIA-99,
>>DOXIA-104,...] (and while your at it, tell me how to fix them without
>>affecting backwards compatibility :) ).
>>
>>It has been mentioned and dicussed a few times on this list, that we
>>intent to stabilize the API with the first beta release. It's at least
>>half a year ago now that I drew up the roadmap for it, and basically all
>>the bug-fix issues scheduled for beta-1 are potentially going to affect
>>backwards compatibility. [1]
>>
>>If people are not happy with that then let's put out alpha-10 as
>>1.0-final and call beta-1 2.0-alpha-30-SNAPSHOT. I'd feel uneasy voting
>>for the release, and my fear is that nothing will happen anymore after
>>that, but fine, if that is what makes people happy.
>>
>>I have tried to collect input on some of the issues on confluence [2]
>>and on this list. It is discouraging to see how people are ignoring such
>>discussions, but as soon as they realize that some changes are going to
>>require some work on their part, they come up with prophetic
>>trivialities ('backward-incompatible changes are baaad') and useless
>>statements ('we've been using it so long', so what?).
>>
>>Don't get me wrong, I'm just trying to get some constructive input,
>>because honestly, I would need it! :)
>>
>>-Lukas
>>
>>[1]
>>http://jira.codehaus.org/browse/DOXIA?report=com.atlassian.jira.plugin.system.project:roadmap-panel
>>[2] http://docs.codehaus.org/display/DOXIA/Home
>>
>>
>>
>>>On 10/21/07, Dennis Lundberg <dennisl@apache.org> wrote:
>>>
>>>
>>>>Lukas has continued work on trunk for beta-1 which includes changes to
>>>>the api. These changes will *not* be in the doxia release that I plan do
>>>>shortly. This release, dubbed alpha-10, is a bug-fix release for
>>>>alpha-9. Alpha-9 has some bugs that makes it unusable for the site- and
>>>>project-info-reports-plugin.
>>>>
>>>>So the plan is to do a quick alpha-10 of doxia followed by releases of
>>>>site-plugin and project-info-reports-plugin.
>>>>
>>>>Carlos Sanchez wrote:
>>>>
>>>>
>>>>>why still alpha?
>>>>>
>>>>>On 10/20/07, Dennis Lundberg <dennisl@apache.org> wrote:
>>>>>
>>>>>
>>>>>>Hi
>>>>>>
>>>>>>As you might have seen from the commit messages, I have created branches
>>>>>>in doxia and doxia-sitetools for future alpha releases. The branches
>>>>>>were created from 1.0-alpha-9. On the branches the following stuff
has
>>>>>>been merged in from trunk:
>>>>>>
>>>>>>- DOXIA-156
>>>>>>- DOXIA-161
>>>>>>- The dependency cleanup in the poms
>>>>>>
>>>>>>Do we need anything else?
>>>>>>
>>>>>>--
>>>>>>Dennis Lundberg
>>>>>>
>>>>>
>>>>>
>>>>--
>>>>Dennis Lundberg
>>>>
>>>
>>>
>>>
> 
> 

Mime
View raw message