On 21/10/15 09:55, Marcus wrote: > Am 10/21/2015 02:25 AM, schrieb Dennis E. Hamilton: >> I recently started a download of all of the AOO 4.1.2 rc2 candidates, >> then see how long it was going to take and let it run overnight. In >> the morning, the download had failed to get them all. >> >> What struck me however is how big the downloads are. >> >> There are full pre-built, installable binaries for 41 separate >> languages, and separately-introducible smaller language packs for each >> of them as well. >> >> The pre-built binaries for each single language take 1s gigabyte, in >> six sets: >> >> 4 flavors for Linux, taking 67% >> 1 flavor for MacOS, for 18% >> 1 flavor for Windows (win32 x86), for 15%. >> >> Yet the popularity of pre-built binaries, based on download >> statistics, is inverted, with about 88% for Windows, 9% for MacOS, and >> 3% for everything else. >> >> Now 3.2% (the actual fraction) of 41 million downloads of AOO 4.1.1 is >> still 1.3 million that include those Linux cases. >> >> QUESTION: Considering only pre-built binaries for downloading, my >> question is, when is it time to reduce those that represent inordinate >> demands to the needs for QA, distribution, and support? >> >> EXCLUDED: >> >> 1. This question is not about the relative use of pre-built full >> binaries for each of 41 languages. There are other questions we could >> ask about that. (The language packs for given languages are much >> smaller than the full binaries though, taking about 20MB per flavor.) >> >> 2. This question is not about the source codes and the SDK. It is >> also not about the value of building and testing for different >> platforms. It is only about distributing pre-built binaries. >> >> I suspect there are more details, and better questions, that could be >> raised. >> >> I'm all ears. What are your concerns? What do you see being >> overlooked in this calculus? > > in other words you are asking the question if and how the file sizes of > the install bits can be reduced? And when it is time to do this? Is this > right or have I understood something wrong? > > Thanks > > Marcus > >
no new question and discussion. This came up at the beginning at Apache and before. I remember that Oracle engineers worked on a new installer that had a multilingual install piece and download everything else on demand (or from a disc I think). I believe we all agree that it doesn't make too much sense as it is today and a smarter packaging and installation would be nice. The point is that many users rely on the localized version and can't really start with an English version only and install the necessary language pack. It's all about convenience of an end user product for the majority of our users. A discussion if it make sense or not is useless, I would suggest to start a discussion on how to make it better from a technical perspective and work on a concept. Juergen --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
