Re: [PATCH]: Fix for several getdate.y issues

[Ondřej Vašík]

Ondřej Vašík wrote:
> Jim Meyering wrote:
> > Would you mind separating that into a few separate change sets?
> ...
> I hope I will do the rest of patches ( 1 or 2 more
> are still necessary to cover all the issues I found).

Here is the second set of patches. 

First one (getdate_4) addresses issue with countable dayshift (like
"next yesterday, last tomorrow etc." . I dropped the change of adding
tWEEK_UNIT as this is not necessary for functionality of the changes. 

Second one (getdate_5) addresses signed relative time offset issue. That
was not possible until time zone is specified(only case where it was
allowed was the hybrid section solution). Grammar before the patch
handles any signed number as timezone (as empty o_colon_minutes and
o_merid section handles anything like valid). I added grammar for signed
numbers without time offset unit (which will handle old required
behaviour of o_colon_minutes and o_merid) and added grammar for signed
relative time offset.

date_2 patch is for coreutils and mentions changes in NEWS and adds one
test (getdate_5 patch related).

Total number of shift/reduce conflicts is 32 (which is 4 lower than 36
in the first all-in-one patch - as there is no tWEEK_UNIT token type).
Increase of the shift/reduce conflicts is caused by o_colon_minutes and
o_merid change - which was required for the functionality of the
patch_5. 

Greetings,
         Ondrej Vasik

getdate_4_countabledayshift.patch
Description: Text Data

getdate_5_timerelsignedoffset.patch
Description: Text Data

date_2_TZrelsignedoffsettest.patch
Description: Text Data

signature.asc
Description: Toto je digitálně podepsaná část zprávy