Re: vc-mode permissions problems on NT

[David Abrahams]

Andre Spiegel <address@hidden> writes:

> On Wed, 2003-09-10 at 15:25, David Abrahams wrote:
>
>> In vc-cvs-state-heuristic, (file-attributes file) yields:
>> 
>>    (nil 1 5 5 (16223 9039) (15707 62723) (15938 41615) 199 "-r--r--r--" nil 
>> 17944 (48170 . 6007))
>> 
>> 
>> (equal checkout-time lastmod) yields:
>> 
>>        nil
>
> Hmm, I need some other information as well.  In addition to the above,
> what is checkout-time for this particular file?  

0

> What time does
> CVS/Entries contain, in plain text?  

...
D/tuple////
D/type_traits////
D/utility////
/index.html/1.2/Thu Aug 15 18:37:55 2002//
D/variant////
D/iterator////
...

> What modification time does the shell report

ls -l (cygwin) sez:

   -r--r--r--    1 dave     None          199 Aug 15  2002 
c:/boost/libs/index.html

and DIR (DOS) sez:

    last written:       08/15/2002  02:37 PM
    last accessed:      09/10/2003  11:44 AM
    created:            02/06/2003  01:59 PM

> and does it agree with
> CVS/Entries

Appears to.

> file-attributes

Currently:

  (nil 1 5 5 (16223 18147) (15707 62723) (15938 41615) 199 "-r--r--r--" nil 
17944 (48170 . 6007))

Does that agree?  I don't know.

> , and checkout-time?  

I don't know how to interpret that number either.

> If CVS/Entries and
> the modification time as reported by the shell agree, then somewhere
> an error must be introduced, so that the above comparison fails.  We
> need to find out where that happens.
>
>> vc-cvs-state doesn't seem to be getting invoked.
>
> That's ok when you find the file.  It should be invoked when you do 
> C-x v v

Nope.

> and if you set vc-[cvs-]stay-local to nil, then it should do an
> actual "cvs status", and thus get the right state of the file, even if
> it guessed wrong in vc-cvs-state-heuristic before.

That works, though it doesn't invoke vc-cvs-state.  But then if I do
`C-x v u' the file doesn't go back to being read-only :(.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com