RE: New push functions

2015-07-21 Thread Patrick Monnerat
Kamil Dudka wrote: > Thanks for the explanation! It makes sense to me now. I was not aware of the VERSIONINFO -> SONAME mapping magic. You're welcome :-) >> On each platform, soname is monotonically increasing, but this is more >> often on iSeries. As a consequence, a linux soname may skip ma

Re: New push functions

2015-07-21 Thread Kamil Dudka
On Monday 20 July 2015 19:28:30 Patrick Monnerat wrote: > The same discussion occurs each time there is some reason to change > VERSIONINFO. > > The answer is: an soname bump always causes the linked programs to be > rebuilt. However, we must talk about VERSIONINFO bump rather than SONAME > bump (

RE: New push functions

2015-07-20 Thread Patrick Monnerat
Kamil Dudka wrote: >> This won't change the final SONAME on platforms which support what you >> described, because they use C-A for SONAME. But others (i.e.: iSeries) >> will bump (they use C alone). > What is the actual motivation to do so? >If applications and other libraries are built agains