svn commit: r408262 - head/editors/uemacs

[Bryan Drewery]

On 2/5/2016 12:04 PM, John Marino wrote:
> On 2/5/2016 8:58 PM, Bryan Drewery wrote:
>> On 2/5/2016 11:52 AM, John Marino wrote:
>>> Libtermcap does not exist in ports, it's not in devel/ncurses.
>>> If one wants doesn't want to use base, the port can't be built.  The
>>> obvious solution is make everythink link with libncurses so both base
>>> and ports options are supported.
>>>
>> I assume this is from privatizing libraries on DragonFly. This is a
>> great effort then as I hope FreeBSD does the same some day. Thanks!
>>
> 
> I was hoping this would be seen as globally beneficial.
> Yes, the library privatization exposed all the linkage to curses,
> termlib, termcap, etc. by breaking the port and is allowing us to
> standardize everything libncurses.
> 
> D5129 is related.
> 
> We're a bit of a mine canary right now, but there's only another 100 or
> so broken ports so the effort is coming to an end.  (we can't fix ports
> that don't build on DF though, obviously)
> 
>> Ok. The massive symlinking to ncurses in base bothers me as it creates
>> issues for build dependencies for me. I have considered trying to
>> eliminate them but haven't gone forward with it since it would break too
>> much.
> 
> I didn't understand this though.  For the most part USES=ncurses is a
> no-op in FreeBSD, isn't it?  If termcap/termlib/curses are all symlinked
> to ncurses, wouldn't linking directly to ncurses be even better?
> Maybing I'm missing the point of this "build dependency" comment.
> 

It's about LIBADD in base. Using LIBADD+=termcap rather than
LIBADD+=ncurses causes issues for some automated testing for overlinking
I am working on.


-- 
Regards,
Bryan Drewery

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 473 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freebsd.org/pipermail/svn-ports-all/attachments/20160205/8efd46df/attachment.sig>