On Mon, Feb 11, 2013 at 6:46 PM, Rob Weir <robw...@apache.org> wrote: > On Mon, Feb 11, 2013 at 6:03 PM, Andrea Pescetti <pesce...@apache.org> wrote: >> On 11/02/2013 Hagar Delest wrote: >>> >>> No real problem with reinstalling extensions after a major upgrade, I've >>> done that too. >>> But there is a difference between the mere inconvenience of reinstalling >>> extensions and losing them completely (waiting that someone dare update >>> them). >> >> >> The real issue is here indeed. Reinstalling won't be perceived as a big >> problem. But the fact that reinstalling the same extension won't work will >> be a problem. >> >> Most of the extensions hosted on extensions.openoffice.org won't be updated, >> and extensions.openoffice.org does not support filtering by version (and >> anyway the information would be missing in current releases). The top five >> extensions at >> http://extensions.openoffice.org/en/most_popular >> total 1 million downloads per year, which could give some backing to the >> nightmare support scenario Hagar envisions. >> > > So you are assuming that the authors of the top extensions will not > update their extensions? Is that a reasonable assumption? Why > wouldn't they update? >
Answering my own question: the top downloaded extension is the PDF Importer, and that has Oracle listed as the owner. If this extension is actually unmaintained, then it would be broken, yes. Of course, if it is unmaintained, then it is in a very fragile position now, and almost anything can break it, from AOO changes, to platform changes to changes in dependent libraries. I'd say: if we don't have a plan for getting such extensions maintained then we should already write them off as broken. They will fall over when the wind blows. It is only a matter of time. So my question then is: are there any top maintained extensions where the author would not adapt it to the proposed AOO 4.0 changes? If this is the case, what is their concern? They don't like the change from a technical perspective? They don't have time to make the change? Something else? -Rob > I agree that if we accepted that assumption then this looks like a bad > change. But I do wonder about the validity of that assumption. > > -Rob > >> Ariel posted to the API list saying that the two reasonable options in his >> opinion are either to keep or revert his entire change (Hagar, please note >> that Ariel asked not to start a discussion here and now, and mentioned he >> cannot be responsive at the moment; anyway...). But if there is a way, even >> using redundant code, to still support the old and new toolbar handling this >> would be very useful to end-users. From the FOSDEM talks I understood there >> could the possibility to still support both mechanisms (and of course, warn >> users when the "deprecated" one is used). >> >> Regards, >> Andrea.