[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] rpm



On Fri, 19 Dec 2014 22:38:00 , Henk Robbers <h.robbers@chello.nl> wrote:
> Peter Slegg schreef op 12/19/14, 9:29 PM:
>
> > Then each would load it's own copy of the libs so there would be
> > A, B, B, C and C in memory. Each app can call it's own copy just
> > like static linked versions.
>
> You seem to forget that most programs only need parts of libraries.
> It could be possible that your programs use different parts of
> each library.
> The memory inefficiency of static linking depends on the size
> of the part of the library used by all programs.
>
> Static linking might not be as inefficient as most people think.
> Besides that its concept is rather uncomplicated and a lot more
> safe than dynamic linking with suspicious libraries
> each time a program is run.
>
>
> Just a few cents.
>


I have often read comments about static linking complaining that an entire
library is linked, even though the app only uses part of it, resulting
in fat apps.

I got the impression that gcc couldn't link part of a lib ?


However, it's not the inefficiency of statics that is the problem, it is
the need to rebuild everything that uses each updated lib.


Even if a small number of libs are updated each month then it becomes
a never ending task to update the existing apps let alone port new ones.

Add to that updates to the apps and the problem becomes insurmountable
for a small number of developers.

I suppose this was a major driver for other platforms to use loadable libs.


If Mark's idea of an automated builder is achieved then it would help to
build the latest releases of a lot of applications but it does sound
complex.

Which solution might be easier in the long run ?

Peter