[cfr] patch to clean up old Linux ports

[Eitan Adler]

On 9 March 2013 16:13, René Ladan <rene at freebsd.org> wrote:
> On 09-03-2013 14:45, Boris Samorodov wrote:
>> 09.03.2013 17:14, Alexander Leidinger пишет:
>>> On Sat, 09 Mar 2013 16:36:03 +0400
>>> Boris Samorodov <bsam at passap.ru> wrote:
>>>> 09.03.2013 15:16, Alexander Leidinger пишет:
>>>>
>>>>> The EoL announcement made it clear that ports need to be marked
>>>>> broken if they don't work on 7, so it means the generic ports
>>>>> framework has no hard "doesn't work" (yet).
>>>>
>>>> As I understand the announcement, those "ports should be marked
>>>> broken", etc. should be done at RELENG_7_EOL tag. Otherwise there is
>>>> no sense at EOL itself.
>>>
>>> BROKEN is used to announce as soon as possible that it will not work,
>>> whereas e.g. a compile error on 7 could manifest it self after a long
>>> time of compiling something.
>>>
>>> Think also about those people which don't know that 7 is EoL, but still
>>> run portsnap. At one point they may want to install a port and then it
>>> fails. If there's no message what's wrong (the system needs to be
>>> updated), they may spend a lot of time to search the cause of the
>>> problem. With a little helpful message they know directly.
>>
>> I agree that a message (well, BROKEN or something else) should be
>> used to inform a user. But that may be done via one check/file.
>> Be it at bsd.ports.mk, bsd.linux.mk, etc. Why should HEAD track
>> individual ports for 7.x after EOL? And when should 7.x actually
>> be cleaned fro the portstree? There is no any other date for 7.x.
>>
>> OK, for those who continue use 7.x RELENG_7_EOL has been created.
>> And those ports committers who are interested in ports for 7.x
>> may use portstree with that particular tag. As well as those
>> users who continue to use FreeBSD 7.x.
>>
> I strongly attend to agree with Boris here. If we want to continue
> warning 7.X users for a while (1,6,12 months?) then it should both be
> much clearer and easier to just put a conditional IGNORE in bsd.port.mk
> than in thousands of individual (not only Linux) ports.

+1.  The place to protect against user error is not in the Linux
specific portion of the tree.

-- 
Eitan Adler