Nearly 4 years ago when I started writing the first lines of OIIO, one of the 
very first things I did is assemble this side project "external", that would 
gather and allegedly make it easy to build, all the dependent packages we would 
need.  It was trying to solve two problems: (1) allow you to separately build 
all the dependencies and store them someplace private (for example, by somebody 
without root access, or who couldn't or wouldn't run a package installer, or 
who didn't want to overwrite their system libs); (2) allow all OIIO devs to use 
exactly the same versions of the dependencies, making development and debugging 
easy.

Of course, it also created new problems: it's finicky, lots of people have 
trouble compiling it, and it's extra work for us to maintain.

In fact, we really don't maintain it any more at all.  I, myself, just use 
Macports or whatever Linux package manager I have on hand to install the 
dependencies (or, at work, I must rely exclusively on the pre-installed ones 
anyway because it's more important to develop on the same libs that I need to 
deploy on).  Almost the only contact I have with "external" is when people 
complain that they can't get it to work.

So, I put it to all of you:

* Does anybody still have the need to separately compile the dependencies 
rather than use your system libs (and package manager)?

* Has anybody ever depended on the fact that the "external" packages were 
locked down and available to all developers?  (I know I haven't, as it turned 
out.)

* Is there any point to keeping the external project going at all, or at this 
point should we just delete it and replace with the list of what packages are 
needed and where to get them from?

Put another way, please speak up if you can think of any reason why we 
shouldn't just give up and delete the "externals" project.


--
Larry Gritz
[email protected]


_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to