incubator-ooo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rob Weir <robw...@apache.org>
Subject Re: LibreOffice relicensing efforts
Date Thu, 24 May 2012 12:02:50 GMT
On Thu, May 24, 2012 at 12:28 AM, Michael Meeks <michael.meeks@suse.com> wrote:
>
> On Wed, 2012-05-23 at 22:51 -0400, Rob Weir wrote:
>> There is very little code of value in AOO that can simply be copied
>> as-is into LO and then never touched again.   Typically the code will
>> need to be modified when initially merged into LO.  But then, as bugs
>> are fixed or the feature is enhanced in AOO, LO will want to merge
>> these patches in again.  And what if then LO has its own unique bug
>> fixes or enhancements to the feature taken from AOO?   Greater merge
>> complexity, maintenance of separate change sets, etc.
>
>        Indeed - it is, in part, for this reason that we make explicit in the
> FAQ that getting your code into AOO is no short-cut to getting it into
> both products.
>
>> The cost of such merges, in terms of developer effort, is significant,
>> as well as being error prone.   For any non-trivial feature it is
>> usually far simpler, and cheaper to "push the patch upstream".
>
>        That really depends on the feature and how it is implemented. Many
> features are rather de-coupled, and/or no equivalent exists on the other
> side. It also depends on whether the Apache project produce nice, clear,
> granular commits (I am assuming you will) :-)
>
>        As for 'upstream', I don't see AOO as an up-stream, but a contributor
> of some volume of changes - no doubt there is lots of mileage in arguing
> about that :-)
>
>>  If they do that, and their patches are accepted into AOO, then they
>> reduce their merge expense.    If they don't do that, they quickly get
>> involved in a costly merge hell.
>
>        Lets see how it goes; this idea that it is impossible to maintain a
> branch and pull code across is rather belied by the existence of a
> five-year closed branch that did exactly that from IBM; and an even
> longer running version of go-oo that worked rather well. Indeed - our
> biggest merge problems were typically associated with code that we had
> contributed to Sun.
>

I said "costly".  I didn't say "impossible".  But it is interesting to
note that Novell and IBM both pushed their patches upstream when given
the opportunity to do so.

>        ATB,
>
>                Michael.
>
> --
> michael.meeks@suse.com  <><, Pseudo Engineer, itinerant idiot
>

Mime
View raw message