[CRIU] [PATCH 0/2] Makefiles: fix deps checking

Dmitry Safonov 0x7f454c46 at gmail.com
Mon Jan 9 07:37:50 PST 2017


2017-01-09 18:14 GMT+03:00 Kir Kolyshkin <kir at virtuozzo.com>:
> On 01/09/2017 12:21 AM, Dmitry Safonov wrote:
>>
>> 2017-01-08 2:52 GMT+03:00 Kir Kolyshkin <kir at openvz.org>:
>>>
>>> A big problem with using sub-makes is the dependency information
>>> is only available to a sub-make, so if we want deps to be checked
>>> when we run make (a natural thing to do, isn't it?), we need to
>>> call all the sub-makes unconditionally.
>>>
>>> I am not sure if a better alternative exist.
>>
>> There is alternative, but it's not an easy-fix: Non-recursive-make.
>
>
> I surely thought about this. The problem in our case is we want
> (or so I think) the sub-packages like libsoccr or compel to be
> autonomous, so we want their Makefiles to be self-sufficient.

It can live with non-recursive, the only problem (and it's big) is
that it will need reworking nmk engine - or even dropping it off.

> In fact this is not the case at the moment, but as I understand
> this is Cyrill's intention.
>
>
>> Started with article "Recursive Make Considered Harmful",
>> which describes all these problems:
>> http://aegis.sourceforge.net/auug97.pdf
>>
>> Reviewed-by: Dmitry Safonov <dsafonov at virtuozzo.com>
>>
>>> PS there is the same problem with lib/c/* but I was not able to fix it
>>> in such a simple way as these two cases, so "leaving it as an exercise
>>> to the reader".
>>>
>>> Kir Kolyshkin (2):
>>>    criu/Makefile: fix rebuilding criu/pie/pie.lib.a
>>>    criu/Makefile: fix criu/arch/* rebuild
>>>
>>>   criu/Makefile | 6 ++++--
>>>   1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>



-- 
             Dmitry


More information about the CRIU mailing list