commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Phil Steitz <phil.ste...@gmail.com>
Subject Re: [DISCUSS] Mission Statement for Commons...
Date Sun, 06 Oct 2013 20:29:56 GMT


> On Oct 6, 2013, at 1:11 PM, Oliver Heger <oliver.heger@oliver-heger.de> wrote:
> 
> Hi Christian,
> 
> Am 06.10.2013 21:44, schrieb Christian Grobmeier:
>> James,
>> 
>> thank you.
>> 
>> I believe Commons is in a bad shape.
>> 
>> Look at Commons Collections. Before 4 years somebody
>> said Guava is more modern, he his answer seems to be widely accepted.
>> http://stackoverflow.com/a/1444467/690771
>> This guy said we have no generics. What did we do in the past 4 years?
>> http://search.maven.org/#search%7Cgav%7C1%7Cg%3A%22commons-collections%22%20AND%20a%3A%22commons-collections%22
>> 
>> 
>> Nothing. At least nothing visible. Its fine we have a beta. I wonder why
>> we haven't managed
>> to officially release this? The last release is from 2008.
>> 
>> I did consider to put my JSON component to Commons. But I didn't.
>> Reason: I really need the component
>> and I calculated how long it would take to release it here. I thought,
>> it's not helping me
>> to develop it here. I simply don't have the time.
>> 
>> I thought a long while about it.
>> 
>> We had discussions like: do we really need Generics? It works without!
>> Do we really drop an outdated JDK? We might have users
>> who run JDK 1.3! And so on. Finally this led us to the situation where
>> almost all of our users seem to have JDK 1.3 and
>> are not interested in generics - in 2013. The users who want modern
>> features go to Guava. We maintain legacy code. And seriously, a lot of
>> code works without generics. This is no reason to not include them.
>> 
>> We discuss magic strings in the sandbox. Why? We don't need to discuss
>> that. Before we release we can simply check Sonar. Safe the time to
>> discuss. Fix it or leave it to Sonar to report it.
>> 
>> We seem to think perfect documentation is more valuable then quick
>> releases. Is it?
>> 
>> We seem to be proud of our build. I am not. It's complex. It's no fun.
>> Releases should be do-able in a short time, maybe
>> even automated.
>> 
>> It is so sad that lot of good features like Collections with Generics
>> were blocked such a long time or drowning in discussions.
>> 
>> I agree we should support old users. But if we don't have the manpower,
>> we can't support them. They need to accept we are moving on. We are
>> blocked with our backwards compatible ideas and innovation is far away.
>> 
>> When I spoke to young developers about Commons they asked me if it still
>> exists. Nobody of them is interested in our community.
>> 
>> For the mission statement I would wish to see things like that:
>> 
>> Commons Components…
>> 
>> …are released quickly and often
>> …do use modern JDKs and support old JDKs only as long as they are
>> supported by Oracle
>> …we make use of modern Java features when they are available
>> …can be easily released
>> …can be released without having 100% perfect documentation or perfect
>> implementations
>> …are build by Community who wants to learn, experiment and create new
>> features more than by Community which wants to be backwards compatible
>> for a long time
>> …are allowed to release major versions with api breaks as they want
>> 
>> Cheers
>> Christian
> I agree with many of your points. Another example is [csv] which is
> about to be released for ages. Here, I think, the main impediment is
> that we try to come up with a *perfect* API because due to our rules of
> backwards compatibility it is so difficult to correct any mistakes later.
> 
> I still think that backwards compatibility is very important, but we
> really should define a process which allows us to experiment with new APIs.
> 
> As a suggestion to improve this situation, could we agree on an alpha
> release process allowing us to push releases with the aim of getting
> community feedback? Where we explicitly state that incompatible changes
> are possible (and likely)?
> 
> We did something similar with [collections] 4, but there were many
> limitations (the release was not allowed to be uploaded to Maven central
> for instance). If we did such experimental releases more often, there
> would hopefully not be the fear of defining a broken API, and we would
> see more releases.

+1 let's agree on how to do alphas.

Phil
> 
> Oliver
> 
>> 
>>> On 6 Oct 2013, at 20:30, James Carman wrote:
>>> 
>>> All,
>>> 
>>> The Apache Commons project seems to be languishing as of late and we
>>> need some rejuvenation.  Perhaps we should try to define our mission
>>> as a project.  What are our goals?  What do we want to accomplish?
>>> Who are our users/customers?  What non-functional qualities do we want
>>> our software to exhibit?  How do we want to conduct ourselves?  How
>>> often do we want to do releases?  What else?
>>> 
>>> Sincerely,
>>> 
>>> James
>>> 
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
>>> For additional commands, e-mail: dev-help@commons.apache.org
>> 
>> 
>> ---
>> http://www.grobmeier.de
>> @grobmeier
>> GPG: 0xA5CC90DB
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
>> For additional commands, e-mail: dev-help@commons.apache.org
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
> For additional commands, e-mail: dev-help@commons.apache.org
> 

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message