Openssl: built-in or ports?

[Rainer Müller]

On 2009-04-03 04:40, Ryan Schmidt wrote:
> Here's an idea: we could make a variable mirror.only, being similar  
> to the existing variable extract.only. If not all files may be  
> mirrored, you can list the ones that may be mirrored in mirror.only,  
> or set mirror.only to empty if none may be mirrored.

Okay, this allows to set mirroring per distfile. But using
  mirror.only
to prevent mirroring looks quite strange to me.

> That doesn't address binary files, since I don't know what filenames  
> our binaries would have. And I think a simple yes/no flag would be  
> more appropriate for whether binaries may be distributed, since it  
> could vary by variant, so it would be most simple to be able to set  
> that flag in the particular variant. e.g. for freetype:

Binary files are out of scope here for mirroring as they do not exist at
the time of writing the Portfile. And a binary file is not a distfile.

> variant bytecode description {Build bytecode interpreter into the  
> TrueType driver} {
> 	create_binaries no
> 	...
> }

What about
  package.create no
The code for creating binary packages lives in package1.0. This could
both be set from the main Portfile or from variants. Default being yes.
Maybe there will be other options necessary for binaries we don't know
yet, so having a package.* namespace sounds like a good idea to me.

> I don't like "mirror.restricted yes" to restrict mirroring; I'd  
> rather make it a positive statement, like "mirror.allowed no" or  
> "distfiles.mirror no". "mirror.restricted yes" feels like having a  
> checkbox that says "[X] Don't mirror" (read: "Yes, don't mirror");  
> checkboxes (or other Booleans) with negative wording are confusing.  
> Simper: "[ ] Mirror"

Okay, I concur with that :-)

Rainer