[vorbis-dev] ao hard codes -ldl in configure tests

[Michael Smith]

At 02:54 AM 11/20/00 -0800, you wrote:
>
>
>  In my opinion, using dlopen in libao was a bad choice in the first
place.  Loadable modules make sense in a few cases:
>
>- When you don't want to release source to the code doing the loading but
still want to allow users to extend it.  Obviously not a concern here.
>- When there are zillions of potential plugins (image filters, for
example).  Obviously not a concern here.
>- When it is too difficult/time consuming/costly for the average developer
to build the system that would do the loading.  Also obviously not a
concern here.
>
>  libao is a _tiny_ piece of code that really does very little and is just
being needlessly complicated by the dlopen() usage.  If I had to guess, I'd
say that someone was interested in learning dlopen() and choose to do it
here, but I think this is a bad design decision given the criteria above.

You missed out the fourth reason:
  When you want to be able to compile plugins that require certain
libraries, and have the rest of the thing work even if that particular
library isn't present. This is important for people that don't compile it
themselves. 

If we want esd support (puke!) in the binaries, then we either need
dynamically loadable modules, or we need to make it REQUIRE esd. Same for
alsa, and any other things that get added in the future.

It does make things overly complex, and makes the otherwise portable libao
no longer so, but it is useful. 

Michael

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.