[ENH] BetterFileStamp

Jan Theodore Galkowski algebraist at salonmember.com
Sun Oct 24 15:53:17 UTC 1999


On Sat, 23 Oct 1999 15:44:49   Jason McVay wrote:
>stephen--
>
>the reasons for not using the ISO 8601 for filestamping are two-(or more)
>fold.
>1) some OSs have relatively short filenames (the mac has a 32 character
>limit) and the ISO 8601 takes up 15 spaces. granted, only a few characters
>are saved with my method compared to ISO 8601, but every character counts...
>2) more importantly, there is no need for using a standard on the filenames
>since the filestamp _is not used_ within the image. it is _solely_ for the
>reader's eyes outside the image. when the filestamped file is filedIn, the
>filestamp is left out.
>
>i encourage you to implement the ISO 8601 standard 'cause standards are
>always good, but using ISO 8601 is overkill in this instance.

What I don't understand is why the filestamp must
occupy the filename slot in the directory table.
There is, of course, a date-time field on most
systems, but that may not be compatible with 8601 or
anything else.  But it does seem to me that as 
files and things like them become more and more
objects in their own right, there is a need for 
a thing like an object header.

One could associate the object header with the
logical file itself, such as demanding that the
first "line" be devoted to identifying information, 
but that'll have an impact.  The other way to do it
would be Bobrow's 1975 (date of publication in CACM)
idea of hash-linking:  Take an invariant associated
with the file, such as its position in the files'
system, and use it as a key to a new system-wide
hash table.  Associate the object identification
tag with the key in that data structure.  There are
problems with using such a centralized broker in a
distributed environment, but it might make local 
sense.  I guess I should add, too, that Bobrow also
proposed using this for allowing people to annotate
local copies of the directory with their own take on
a thing: One of the proposed applications was layers
of notes on a document, with classifying tags in the
hash table to separate those notes one wanted to keep
private from those one wanted to share with certain
other groups.

---
________________________________________________
Jan Theodore Galkowski  demiourgos at smalltalk.org
home.stny.rr.com/algebraist/         squeak.org/
www.smalltalk.org/



Get your FREE, private e-mail account at http://www.salonmember.com.





More information about the Squeak-dev mailing list