On 19 Jan 2014, at 12:19, Stéphane Ducasse <[email protected]> wrote:
> I guess that once we will use git it will go away. Does anybody know if this > is a correct assmption? yes, is correct :) alll that information can live outside the image without actually lose of information (it will be there, when we need it). > > On 19 Jan 2014, at 10:10, Marcus Denker <[email protected]> wrote: > >> Very very nice! I integrate issue 12679. >> >> Yes, Monticello meta data… we should check this issue: >> >> >> https://pharo.fogbugz.com/f/cases/2711/Monticello-wastes-significant-amounts-of-memory >> >> (but in general it is not good to have meta data of a versioning system in >> the image… it needs to be on disk, and the history on >> disk needs to not by in thousands of zip files…). >> >> On 18 Jan 2014, at 20:38, Pavel Krivanek <[email protected]> wrote: >> >>> Hi, >>> >>> I finally successfully created a script that is able to unload everything >>> *by Monticello* except kernel packages, network, Monticello, Gofer and >>> related packages. I will setup a job when Jenkins will be healthy again. >>> It uses several temporary patches that we need to solve properly: >>> - make NativeBoost optional >>> - make UserManager optional >>> - small super-easy issue >>> https://pharo.fogbugz.com/f/cases/12679/change-package-of-ClassTrait-soleInstance >>> >>> The size of resultant image is about 6.6 MB, 4.9 MB without MC data and >>> caches. BTW in the full image Monticello data have about 5MB. The image has >>> no Undeclared nor obsolete classes and is really able to load something >>> using Gofer. >>> >>> With some care we can make it usable base for the next remodularization >>> effort. >>> >>> Cheers, >>> -- Pavel >>> >> >
