home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
icon
/
newsgrp
/
group94a.txt
/
000069_icon-group-sender _Thu Mar 3 10:15:32 1994.msg
< prev
next >
Wrap
Internet Message Format
|
1994-08-19
|
3KB
Received: by cheltenham.cs.arizona.edu; Thu, 3 Mar 1994 12:35:39 MST
Message-Id: <m0pcFDg-000NAjC@epic.truevision.com>
Date: 3 Mar 1994 10:15:32 -0500
From: "Paul Scherf" <paul_scherf@gatormail-m.truevision.com>
Subject: RE: file dates
To: icon-group@cs.arizona.edu
Status: R
Errors-To: icon-group-errors@cs.arizona.edu
I think this sounds like a great chance for defining a library interface
(collection of functions), that could be supported by many of the environments
in which ICON is used. My opinion is that new language keywords are a bit much.
I guess that makes me against adding keywords.
C (i.e. past history (-: ) effectively defines two collections of functions for
dealing with time. One collection deals with calendar dates and times of day
(e.g. "What was the time and date of lunch one week from now?"). The other
collection deals with stopwatch time (for scientific timing of events). Leap
seconds and leap days (these are decided by astronomers and relative positions
of celestial bodies) and calendars that vary with time (different number of
months in different years, these appear to be decided by politicians and
religous leaders), make these two kinds of time difficult to integrate.
Since ICON is often implemented in C, maybe an ICON interface to the C
libraries would be useful or even appropriate.
_______________________________________________________________________________
In article <5a-oml5.mengarini@delphi.com>,
Will Mengarini <mengarini@delphi.com> wrote:
>but I think this is a very good point that it'd be useful
>for the language to define keywords or functions for things
>that can reasonably be expected to be present on most or all
>implementations.
You have to way that against increasing the size of the language
(bloating or creeping featurism). If there is a way to do it without
adding a language feature (which you have already stated is possible
under MS-DOS, and I can think of ways to do it under Unix), then I'd
lean against adding it as a feature.
>Can anybody name an Icon implementation for a system where
>files are created & stored in a way that doesn't keep track
>of both creation date & time, & last-change date & time?
That's still not enough to make it useful. At a minimum, you would
like to be able to compare dates (eg: is date1 later than date2).
You'd probably wish to know the amount of time between two dates, and a
way to compare it to the current date. This would be highly machine
dependent (does the machine keep track of timestamps to the nearest
microsecond or to the nearest minute). You'd probably want a standard
textual representation to make it easy to create new dates to compare
against. But if you used something like &dateline as the standard
representation, should you provide functions to fix the redundant
fields (the day of the week can be calculated from the month day year
combination, making it redundant)? Would you add functions to add or
subtract time to a given date? Etc., etc.