[illumos-Advocates] RTI: 1168 want a libtermcap that does not force libcurses

[Gordon Ross]

On Fri, Jul 15, 2011 at 4:49 PM, Richard Lowe <richlowe at richlowe.net> wrote:
> My assumption would be that it failed for a user doing the "other"
> type of build to you, if you're build was full (which I hope it was!)
> it probably failed on an incremental.  Given that the noise diffs you
> pasted was the noise going away, I'm assuming you tested this
> successfully now?

Yeah.

> If so, I guess I'm happy.
>
> -- Rich
>
>
>
> On Fri, Jul 15, 2011 at 16:34, Gordon Ross <gordon.w.ross at gmail.com> wrote:
>> The diffs correct an error (well, noise) in a full clean build:
>>  ==== Build noise differences (DEBUG) ====
>> 161,162d160
>> < ln: cannot create ./llib-ltermcap: File exists
>> < ln: cannot create ./llib-ltermlib: File exists
>>
>> These were for the llib-ltermcap and llib-ltermlib
>> files under $(ARCH64) directories, which were
>> never supposed to be there.  My bug here was
>> deleting only part of the Targetdirs stuff that
>> used to make those links.  This completes it.
>>
>> Gordon
>>
>> On Fri, Jul 15, 2011 at 4:11 PM, Richard Lowe <richlowe at richlowe.net> wrote:
>>> These diffs look incorrect to me.  I didn't think you actually had
>>> touched termlib.
>>>
>>> -- Rich
>>>
>>>
>>>
>>> On Fri, Jul 15, 2011 at 14:56, Gordon Ross <gordon.w.ross at gmail.com> wrote:
>>>> Gah!  Looks like I somehow missed a change in $SRC/Targetdirs
>>>>
>>>> diff -r 71e59c2d8715 usr/src/Targetdirs
>>>> --- a/usr/src/Targetdirs        Fri Jul 15 10:20:38 2011 -0400
>>>> +++ b/usr/src/Targetdirs        Fri Jul 15 14:54:11 2011 -0400
>>>> @@ -1902,9 +1902,7 @@
>>>>        /usr/lib/$(MACH64)/llib-lsendfile.ln \
>>>>        /usr/lib/$(MACH64)/llib-lsocket.ln \
>>>>        /usr/lib/$(MACH64)/llib-lsysevent.ln \
>>>> -       /usr/lib/$(MACH64)/llib-ltermcap \
>>>>        /usr/lib/$(MACH64)/llib-ltermcap.ln \
>>>> -       /usr/lib/$(MACH64)/llib-ltermlib \
>>>>        /usr/lib/$(MACH64)/llib-ltermlib.ln \
>>>>        /usr/lib/$(MACH64)/llib-lthread.ln \
>>>>        /usr/lib/$(MACH64)/llib-lthread_db.ln \
>>>>
>>>> It broke someone else's build, but not mine.
>>>> Not sure why...
>>>>
>>>> I propose to push the above with the same issue #
>>>> and (fix Targetdirs), if that's OK with y'all.
>>>> (After another full, clean build finishes.)
>>>>
>>>> Gordon
>>>>
>>>>
>>>> On Thu, Jul 14, 2011 at 11:30 PM, Garrett D'Amore <garrett at damore.org> wrote:
>>>>> approved
>>>>>
>>>>>  -- Garrett D'Amore
>>>>>
>>>>> On Jul 14, 2011, at 5:13 PM, Gordon Ross <gordon.w.ross at gmail.com> wrote:
>>>>>
>>>>>> As discussed on illumos-dev, re. issue 514, etc. this introduces a
>>>>>> new "filter" library allowing programs and libraries (i.e. readline)
>>>>>> to express a dependency on termcap functions without forcing a
>>>>>> choice of which curses library they use, if any.
>>>>>>
>>>>>> Final webrev:
>>>>>>  http://www.yalms.org/cr/libtermcap/
>>>>>>
>>>>>> outgoing, pbchk, mail_msg (attached)
>>>>>>
>>>>>> The mail_msg has some noise due to:
>>>>>> https://www.illumos.org/issues/1078
>>>>>> but I don't see any real problems there.
>>>>>>
>>>>>> Tested using a special build of libreadline with a helper function
>>>>>> that returns the address of the tgetent function it's linked with.
>>>>>> I called that with two test programs, one linked with libreadline
>>>>>> and libncurses, the other linked with libcurses, ad verified that
>>>>>> the correct tgetent function (i.e. ncurses - the one linked in by
>>>>>> the test program) was called from inside libreadline.
>>>>>>
>>>>>> Also, Julian Wiesener <jw at vtoc.de> did a full SFW build and
>>>>>> checked some (possibly) affected programs build there.
>>>>>>
>>>>>> Gordon
>>>>
>>>> _______________________________________________
>>>> Advocates mailing list
>>>> Advocates at lists.illumos.org
>>>> http://lists.illumos.org/m/listinfo/advocates
>>>>
>>>
>>
>