home *** CD-ROM | disk | FTP | other *** search
- Path: mail2news.demon.co.uk!oxigen.demon.co.uk
- From: Dirk Wessels <dirk@oxigen.demon.co.uk>
- Newsgroups: comp.lang.smalltalk,comp.object,comp.lang.c++,comp.lang.java
- Subject: Re: The Good, the Bad, the Ugly, and the Wicked ...
- Date: Tue, 26 Mar 96 14:38:24 GMT
- Organization: O2
- Message-ID: <827851104snz@oxigen.demon.co.uk>
- References: <31570B8E.5A12@vmark.com>
- Reply-To: dirk@oxigen.demon.co.uk
- X-NNTP-Posting-Host: oxigen.demon.co.uk
- X-Newsreader: Demon Internet Simple News v1.30
- X-Mail2News-Path: oxigen.demon.co.uk
-
- In article <31570B8E.5A12@vmark.com>
- jsutherland@vmark.com "Jeff Sutherland" writes:
-
- >
- > ST C++ OOC Java
- > Flexibility Dynamic Binding 1 2 2 2
- > Dynamic Classes 1 3 1 2
- > Multiple Inheritance 3 2 2 3
- > Roles 2 3 3 1
- > Ease of use Class Libraries 1 3 3 2
- > Learning Curve 1 3 2 1
- > Speed of Development 1 3 2 2
- > Portability 2 3 3 1
- > Support Tools 1 1 3 3
- > Multiple Vendors 2 1 3 1
- > Performance 2 1 3 3
- > Risk Garbage Collection 1 3 3 2
- > Memory Leaks 1 3 1 1
- > Overwriting Memory 1 3 1 1
- > Ready for Prime Time 1 1 2 3
- > TOTAL (low means best) 21 35 34 28
- >
-
- Everybody can make a nice table, here is mine:
- (skipping OOC)
- *= integrated in environment.
- += can be improved in future.
-
- ST C++ OOC Java
- Distributed
- Processing 1 3 1
- Data-management 1 3 2
- Object databases 1 2 3
- Portability 2+ 3 1
- Persistence 1* 2 3
- -----------------------------------------------------------------
- 6 13 10
-
- While Java is build for remote processing, this feature was already
- available for ST, and is still being improved. There are more object
- databases and middleware available for C++, but ST does not need
- separate middleware and its object-databases seem much more mature.
-
- Efficiency
- Speed 2+ 1 3+
- Application size 3+ 2 1
- Source code size 1 2 3
- Tuning 1 2 3
- Risks 1 3 2
- -----------------------------------------------------------------
- 8 10 12
-
- I agree ST is not regarded as very efficient, but because it is very
- dynamic it can be tuned to match almost every requirement.
- The application size can only be reduced with careful packaging.
- If efficiency is your highest priority than you could also consider
- ANSI-C, Fortran, Assembler.
- As already proved for Assembler, future compiler improvements (SELF)
- and hardware changes (Java-processors) may change all these statistics
- considerably.
-
-
- Resources
- Books 3 1 2
- Hype 1 2 3
- Examples 2* 1 3
- 3rd party 2 1* 3
- Programmers 2 1 3
- Analists 1 2 3
- Quality 1 2 3
- Mature 1 1 3
- Class Libraries 1 2 3
- Learning curve 1 3 2
- -----------------------------------------------------------------
- 15 15 28
-
- Development& maintenance
- Time 1 3 2
- Testing 1 3 2
- Debugging 1 3 2
- Tools 1* 2 3+
- Recovery 1 3 2
- Team-support 1 2 3
- Incremental development 1 3 2
- Reuse of constructs (classes/templ) 1 2 3
- Effect of requirement changes 1 3 2
- High level programming 1 3 2
- Low level programming 2 1 3
-
- Stability of standard 1+ 3+ 2+
- -----------------------------------------------------------------
- 13 31 27
-
- As shown by the + signs, with my expectations of future improvements
- (compilers and hardware always get better), Smalltalk may even
- get more ahead and could win on almost every point.
-
- Still offcourse one can give different priorities to each option,
- and come with the conclusion that a different language is better.
-
- Has anybody ever made a table like this with money ($$$) values in them,
- taken from actual project information?
- It may be the only generally accepted measure.
-
-
- Ciao,
-
- Dirk
-
- ____________________________________
- Dirk Wessels
- (dirk@oxigen.demon.co.uk)
-
- 'nothing has happened until observed' -- law of physics.
-