Newsgroups: comp.lang.smalltalk
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornell!travelers.mail.cornell.edu!news.tc.cornell.edu!caen!zip.eecs.umich.edu!newshost.marcam.com!news.mathworks.com!udel!gatech!howland.reston.ans.net!usc!rutgers!uwvax!sinetnews!news.u-tokyo.ac.jp!kappa!tkl.iis.u-tokyo!tklgw!suzuki
From: suzuki@tkl.iis.u-tokyo.ac.jp (Shinji Suzuki)
Subject: Re: Persistence?
In-Reply-To: bfelton@ibm.net's message of 10 Feb 1995 16:07:59 GMT
Message-ID: <SUZUKI.95Feb14211047@yebis.tkl.iis.u-tokyo.ac.jp>
Sender: news@tkl.iis.u-tokyo.ac.jp
Organization: Institute of industrial science, University of Tokyo
References: <3h7uba$qav@tpd.dsccc.com> <3hapmb$2gdp@news-s01.ny.us.ibm.net>
	<3hesc4$htt@picasso.cssc-syd.tansu.com.au>
	<3hg30v$6d7@news-s01.ny.us.ibm.net>
Date: Tue, 14 Feb 1995 12:10:47 GMT
Lines: 23

>In article <3hg30v$6d7@news-s01.ny.us.ibm.net> bfelton@ibm.net writes:

>Persistence is a trivial problem in Smalltalk *for single user applications*.

Aside from issues arising from sharing, there are issues from scale.
The 'dumping' model for implementing persistence does not work well
when the number and the size of objects increase. [Due to the
longer startup time and lack of enough address space.] I believe a
persistent object system should be able to incrementally load
persistent objects into volatile heap. [Not that I dispute against
the idea that Smalltalk environment provides object persistence.]

Recommended Readings for Smalltalkers:
Ted Kaehler and Glenn Krasner, 'LOOM - Large Object Oriented
Memory for Smalltalk-80 Systems', Smalltalk-80 Bis of History, Words of
Advice, Addison Wesley

Antony L. Hosking, J.Eliot B. Moss, 'Object Fault Handling for
Persistent Programming Lanugages: A Performance Evaluation',
Proceedings ACM Conference on Object Oriented Programming Systems,
Languages and Applications, Washington DC, Sep. 1993, pp.288-303

-shinji
