I wrote up the current status in our Wiki

https://cwiki.apache.org/confluence/display/MAVEN/MavenSharedUtils

I've not yet finished the whole class matrix, will do that in the afternoon.

LieGrue,
strub



----- Original Message -----
> From: Mark Struberg <[email protected]>
> To: Maven Developers List <[email protected]>
> Cc: 
> Sent: Thursday, August 30, 2012 12:01 PM
> Subject: Re: Any maven-shared or maven-core release pending?
> 
> 
>>  Shading/delegating is the road to hell, later on it becomes easier to just 
> add 
>>  one method implementation there and the show starts.
> 
> Nope, with shading in the source/class dynamically it will _not_ kept in our 
> SVN! Thus noone could tweak em ;)
> 
> 
>>  Directly using commons-* 
>>  wherever needed should be the preferred way.
> That's the reason why I did NOT add any Base64 nor similar to m-shared-utils 
> yet.
> One counter argument could be binary compatibility. We need some test project 
> with _very_ old plugins to check if we are still fine. 
> If maven-core brings in commons libs as dependencies, then we might clash 
> with 
> very old plugins which need ancient versions of commons as well.
> That kind of stuff is the the reason why framework projects most times shade 
> those classes away into own packages which do not create classpath clashes.
> Even Sun does that with lots of Apache classes which get shipped inside the 
> JRE 
> jars.
> But we will for sure always take the original sources from commons and not 
> maintain them ourselfs!
> 
>>  In short I'm looking for smth. like https://live.gnome.org/GnomeGoals 
> list 
> 
> I already added a table to the javadocs package.html. Not sure if this will 
> remain there, but for now it's fine.
> 
> LieGrue,
> strub
> 
> 
> ----- Original Message -----
>>  From: Aleksandar Kurtakov <[email protected]>
>>  To: Maven Developers List <[email protected]>
>>  Cc: 
>>  Sent: Thursday, August 30, 2012 11:40 AM
>>  Subject: Re: Any maven-shared or maven-core release pending?
>> 
>> 
>> 
>>  ----- Original Message -----
>>>   From: "Mark Struberg" <[email protected]>
>>>   To: "Maven Developers List" <[email protected]>
>>>   Sent: Thursday, August 30, 2012 12:12:49 PM
>>>   Subject: Re: Any maven-shared or maven-core release pending?
>>> 
>>>   yes, that is exactly the goal. E.g. instead of imlementing our own
>>>   Base64 stuff we will either shade in parts or delegate over to
>>>   commons-io.
>> 
>>  Shading/delegating is the road to hell, later on it becomes easier to just 
> add 
>>  one method implementation there and the show starts. Directly using 
> commons-* 
>>  wherever needed should be the preferred way. Not to mention that it will be 
> way 
>>  more familiar for people when they see org.apache.commons and having dealt 
> with 
>>  it instead of org.apache.maven.shared.utils. At least for people like me it 
> will 
>>  not make me move away because of the 48th Base64 implementation I see this 
>>  month. Even if it's simply calling commons-codec, different 
>>  class/method/parameter names make it a different API.
>> 
>>> 
>>>   After the initial switch from p-u to m-s-u we will review all code
>>>   and make a major upgrade and drop obsolete parts.
>>  Doesn't that mean that this shading library will stay forever in 
>>  Maven+plugins deps just like plexus-containers-default 1.0 alpha(5...30)? 
>>  Also is the effort to moving to Apache commons ongoing? 
>>  Is there some page containing the pending conversions, what is the 
> preferred 
>>  replacement, status of core components? 
>>  In short I'm looking for smth. like https://live.gnome.org/GnomeGoals 
> list 
>>  and https://live.gnome.org/GnomeGoals/GSettingsMigration as a detailed 
> example. 
>>  Are you actively looking for external help in this direction? Maybe I can 
>>  promote it to Fedora Java people as we already have done a lot of work to 
>>  streamline the dependencies so it becomes supportable for people building 
> from 
>>  source and this is something that Linux distro guys are doing with or 
> without 
>>  upstream so happening upstream would benefit everyone. But for this to 
> happen 
>>  there needs to be an active upstream developer willing to review/commit 
> patches.
>> 
>>  P.S My colleague Stanislav meant something like that when he wrote about 
> porting 
>>  plugins away of maven-compat, when he wrote to that list sometime ago.
>> 
>>  Regards,
>>  Alexander Kurtakov
>>  Red Hat Eclipse Team
>> 
>>> 
>>>   LieGrue,
>>>   strub
>>> 
>>> 
>>> 
>>> 
>>>   ----- Original Message -----
>>>   > From: Aleksandar Kurtakov <[email protected]>
>>>   > To: Maven Developers List <[email protected]>; Mark 
> Struberg
>>>   > <[email protected]>
>>>   > Cc:
>>>   > Sent: Thursday, August 30, 2012 11:07 AM
>>>   > Subject: Re: Any maven-shared or maven-core release pending?
>>>   > 
>>>   > 
>>>   > 
>>>   > ----- Original Message -----
>>>   >>  From: "Mark Struberg" <[email protected]>
>>>   >>  To: "Maven Dev" <[email protected]>
>>>   >>  Sent: Thursday, August 30, 2012 11:55:13 AM
>>>   >>  Subject: Any maven-shared or maven-core release pending?
>>>   >> 
>>>   >>  Hi folks!
>>>   >> 
>>>   >>  Is there any maven-core or maven-shared release pending?
>>>   >>  I'd like to start with moving all maven stuff from 
>>  plexus-utils
>>>   >>  to
>>>   >>  maven-shared-utils.
>>>   >> 
>>>   >>  The reason is simple:
>>>   >>  1.) It always got argued that we cannot just replace 
> plexus-utils
>>>   >>  because we don't know how many plugins are using it. And 
> 
>>  people
>>>   >>  feared about side effects which could harm stability.
>>>   >>  As we do this now in a whole new module with a new package 
> name
>>>   >>  we
>>>   >>  can really easily upgrade that stuff to proper 2012 style 
> and
>>>   >>  don't
>>>   >>  need to keep em 2005ish anymore ;)
>>>   >> 
>>>   >>  2.) For implementing the incremental build stuff I need a 
> few
>>>   >>  classes
>>>   >>  extended, and the only way we can guarantee this to be 
> reliably
>>>   >>  available is to host this stuff in maven itself (where it 
> was
>>>   >>  originally hosted btw - 80% of the plexus utils code is a 
> fork of
>>>   >>  old ant/jakarta/commons stuff). And since the codehaus repo 
> got
>>>   >>  closed down silently I have no confidence in that stuff 
> anymore.
>>>   >> 
>>>   > Can we hope that stuff that belongs to Apache commons will be 
> used
>>>   > from there or
>>>   > contributed back instead of growing yet another utility library
>>>   > that duplicates
>>>   > commons-[io|lang|codec|beanutils|..]?
>>>   > Java land is already full of such util libraries serving same
>>>   > purposes but with
>>>   > different problems and bugs, so let's try to not do it again.
>>>   > 
>>>   > Regards,
>>>   > Alexander Kurtakov
>>>   > Red Hat Eclipse Team
>>>   > 
>>>   > 
>>>   >> 
>>>   >>  I locally already upgraded a few maven-shared libs without 
> any
>>>   >>  problems. I think we can get this done in 1 week + another 
> week
>>>   >>  of
>>>   >>  cleaning up the obsolete code as Kristian proposed in the 
> other
>>>   >>  mail.
>>>   >> 
>>>   >>  Wdyt? When may I start with the transition?
>>>   >> 
>>>   >>  LieGrue,
>>>   >>  strub
>>>   >> 
>>>   >> 
>>>   >>  
>>  ---------------------------------------------------------------------
>>>   >>  To unsubscribe, e-mail: [email protected]
>>>   >>  For additional commands, e-mail: [email protected]
>>>   >> 
>>>   >> 
>>>   > 
>>> 
>>>   ---------------------------------------------------------------------
>>>   To unsubscribe, e-mail: [email protected]
>>>   For additional commands, e-mail: [email protected]
>>> 
>>> 
>> 
>>  ---------------------------------------------------------------------
>>  To unsubscribe, e-mail: [email protected]
>>  For additional commands, e-mail: [email protected]
>> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to