Bug#163666: debian-policy: Unclear result with [arch] and |
On Sat, Oct 26, 2002 at 12:15:55PM -0500, Manoj Srivastava wrote:
> Hi,
> >>"Michal" == Michal 'hramrach' Suchanek <hramrach_l@centrum.cz> writes:
>
>
> Michal> In section 7.1 the Debain Policy manual says that 'pkg
> Michal> (version) [arch]' is ignored for calculating dependencies.
> Michal> This is not really clear as discussed in bug 160129.
>
> I am missing something; reading the bug report seemed to me to
> be unrelated to pkg (version) [arch] construct; they seemed to be
> talkign about pkg | arch, which is wrong.
no, it was if "pkg [arch] | otherpgk [!arch]" is correct.
It is certainly easier to replace '|' with ',' and avoid investigating
the exact meaning of "pkg [arch] | otherpkg".
But it may be useful in other cases.
>
> Michal> One could think that 'pkg [arch] | otherpkg' is interpreted
> Michal> as 'otherpkg' in !arch case because 'pkg [arch]' is ignored.
>
> OK
>
> Michal> On the other hand, 'pkg [arch]' can be easily implemented in
> Michal> logic as 'arch IMPLIES pkg',
>
> Umm. I am not sure this says what you think it says.
>
> Michal> which is true for !arch and makes the dependency appear
> Michal> 'ignored' in the simple case w/o |. But 'pkg [arch] |
> Michal> otherpkg' would be always true for !arch this way and no
> Michal> dependency on otherpkg ever required.
>
> I am not sure I understand this paragraph.
>
> Michal> I'd appreciate if the documentation described implemented
> Michal> behavior clearly.
>
> If someone suggests alternate working, I'd be happy to include
> it (provided that I understand it ;-)
(arch => pkg) is equal to ( !arch or pkg)
which is true for !arch == true as well as pkg == true
This works for the case without '|'.
This is Marcus' suggested interpretation and it would apply for
building some sort of parser "from bottom" in some logic language
which is probably not the case ;)
Anyway, there is no wording or example in the doc that supports either
interpretation explicitly. At least I could not fing one.
Unless this is clarified in the doc I guess both interpretations are
vaild.
--
Michal Suchanek
hramrach@centrum.cz
Reply to: