Re: [Bug-apl] Fix time inconsistencies

[Juergen Sauermann]

Hi David,

I have improved the fix time handling a little, see SVN 331. As I have
mentioned earlier there will not be full support for function attributes 
like ⎕AT
in GNU APL because:

1. ⎕AT is non-standard and I believe the fix time of a function is not 
too useful.
2. ⎕AT has no way to set some attributes. As a consequence )LOAD and 
)COPY with
   .apl files cannot set the time stamp.
3. )LOAD and )SAVE with .xml files should sort-of work now for ⎕AT, 
except for timezones.
  There is no specification or description saying if the timestamps 
shall be GMT or localtime.
 I personally find it more plausible to use GMT all over the place 
since files are shipped around
 and therefore localtime is ambiguous. On the othe hand, users will 
likely be confused seeing GMT
 instead of localtime.
4. )OUT worked since the exchange format had provisions for timestamps.

/// Jürgen


On 06/17/2014 05:52 AM, David B. Lamkins wrote:
> On Mon, 2014-06-16 at 10:32 -0700, David B. Lamkins wrote:
> > There does not seem to be a case in which fixing a function results in a
> > timestamp other than 0 0 0 0 0 0 0 as a result of 2⎕AT. However, saving
> > a workspace via )OUT does result in *some* functions having a reasonable
> > timestamp while others have their timestamp set to the Linux Epoch (1970
> > 1 1 0 0 0 0).
> >
> > I believe there are two issues:
> >
> > 1) Fixing a function in APL should set the function's timestamp; 2⎕AT
> > should return the timestamp and not 0 0 0 0 0 0 0.
> >
> > 2) )OUT should write the same timestamps that 2⎕AT returns.
> Added:
>
> 3) .xml workspaces should include a function's timestamp as an
> attribute. )LOAD should restore the timestamp.