On Jul 29, 6:31 pm, gavino <·········@gmail.com> wrote:
> http://bc.tech.coop/blog/040128.htmlwas reading here about linux
" he article indicated that the Linux kernel alone was made up of 2.4
million SLOC (mostly C code). If you look at the above table, compare
the Linux 2.4 million SLOC with the 460 thousand SLOC for the
Symbolics Genera Lisp Machine (an OS that was arguably far superior to
Linux). "
--- quoted from the blog article.
Is the Linux's 2.4 million SLOC just the size of kernel? Or includes
device drive?
·········@gmail.com wrote:
> Is the Linux's 2.4 million SLOC just the size of kernel? Or includes
> device drive?
It doesn't make sense to compare this. Maybe there was some useful software
in the Symbolics code, but I doubt that it had drivers for all kind of
hardware like Linux, e.g. Linux supports more than 60 network cards (
http://tinyurl.com/34zclc ) and I assume there are lots of device types the
Symbolics machine not even know, but the Lisp machine has some application
code, which the Linux kernel doesn't have.
You can download the Linux kernel from http://www.kernel.org/ , e.g. the
latest kernel:
ftp://ftp.kernel.org/pub/linux/kernel/v2.6/linux-2.6.22.tar.gz
Then you can measure the number of lines:
find . -exec cat {} \; | wc
8,499,410 29,654,309 244,866,374 (lines, words, chars)
find . -type f -name *.c -exec cat {} \; | wc
5,979,935 19,411,234 16,2966,504
find . -type f -name *.h -exec cat {} \; | wc
1,462,784 6,084,899 52,288,919
I don't know which kernel version the blogger used for his 2,4 million
lines of code. Adding the headers and C files, the Linux kernel has 7,4
million lines of code.
Symbolics Lisp machines were sold with the computer, so maybe it supports
one network card? This means it would need 460,000 * 60 = 28 million lines
of code to support all drivers Linux does. The conclusion: C is nearly 4
times more efficient in code size than Lisp :-)
--
Frank Buss, ··@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de
On Jul 29, 5:08 am, Frank Buss <····@frank-buss.de> wrote:
> ·········@gmail.com wrote:
> > Is the Linux's 2.4 million SLOC just the size of kernel? Or includes
> > device drive?
>
> It doesn't make sense to compare this. Maybe there was some useful software
> in the Symbolics code, but I doubt that it had drivers for all kind of
> hardware like Linux, e.g. Linux supports more than 60 network cards (http://tinyurl.com/34zclc) and I assume there are lots of device types the
> Symbolics machine not even know, but the Lisp machine has some application
> code, which the Linux kernel doesn't have.
>
> You can download the Linux kernel fromhttp://www.kernel.org/, e.g. the
> latest kernel:ftp://ftp.kernel.org/pub/linux/kernel/v2.6/linux-2.6.22.tar.gz
> Then you can measure the number of lines:
>
> find . -exec cat {} \; | wc
> 8,499,410 29,654,309 244,866,374 (lines, words, chars)
>
> find . -type f -name *.c -exec cat {} \; | wc
> 5,979,935 19,411,234 16,2966,504
>
> find . -type f -name *.h -exec cat {} \; | wc
> 1,462,784 6,084,899 52,288,919
>
> I don't know which kernel version the blogger used for his 2,4 million
> lines of code. Adding the headers and C files, the Linux kernel has 7,4
> million lines of code.
>
> Symbolics Lisp machines were sold with the computer, so maybe it supports
> one network card? This means it would need 460,000 * 60 = 28 million lines
> of code to support all drivers Linux does. The conclusion: C is nearly 4
> times more efficient in code size than Lisp :-)
>
> --
> Frank Buss, ····@frank-buss.dehttp://www.frank-buss.de,http://www.it4-systems.de
hm
has anyone ever used a lisp machine as an e commerce server? (aka not
just a workstation)
On Jul 29, 2:54 pm, gavino <·········@gmail.com> wrote:
> has anyone ever used a lisp machine as an e commerce server? (aka not
> just a workstation)
One story I've heard is that back in the 1980s, American Express
developed an expert system (which might still be in use) for approving
credit card transactions. As I understand, AmEx doesn't assign a
specific credit limit to an account; rather, they approve transactions
on a case-by-case basis depending on various factors such as the
account's existing balance, payment history, size of the requested
transaction, and sizes of past transactions. (So e.g. if you have a
history of making large purchases and paying them off promptly,
they'll keep allowing you to do that.)
Anyway, I recall being told that somewhere AmEx had a bank of
Symbolics machines running this expert system, making these
transaction approval decisions in real time. That may not be exactly
what you have in mind, but I think it demonstrates that these machines
were perfectly capable of providing E-commerce services.
-- Scott
In article <·······················@i38g2000prf.googlegroups.com>,
Scott Burson <········@gmail.com> wrote:
> On Jul 29, 2:54 pm, gavino <·········@gmail.com> wrote:
>
> > has anyone ever used a lisp machine as an e commerce server? (aka not
> > just a workstation)
>
> One story I've heard is that back in the 1980s, American Express
> developed an expert system (which might still be in use) for approving
> credit card transactions. As I understand, AmEx doesn't assign a
> specific credit limit to an account; rather, they approve transactions
> on a case-by-case basis depending on various factors such as the
> account's existing balance, payment history, size of the requested
> transaction, and sizes of past transactions. (So e.g. if you have a
> history of making large purchases and paying them off promptly,
> they'll keep allowing you to do that.)
>
> Anyway, I recall being told that somewhere AmEx had a bank of
> Symbolics machines running this expert system, making these
> transaction approval decisions in real time. That may not be exactly
> what you have in mind, but I think it demonstrates that these machines
> were perfectly capable of providing E-commerce services.
>
> -- Scott
That's the 'Authorizer's Assistant'.
Symbolics cites it also in an old brochure:
http://lispm.dyndns.org/symbolics-5.html
This page mentions the app:
http://lispm.dyndns.org/symbolics-5/thumbnails/Genera_11.jpg
--
http://lispm.dyndns.org
In article <·······················@i38g2000prf.googlegroups.com>,
Scott Burson <········@gmail.com> wrote:
> On Jul 29, 2:54 pm, gavino <·········@gmail.com> wrote:
>
> > has anyone ever used a lisp machine as an e commerce server? (aka not
> > just a workstation)
I don't think there has been any new Lisp Machine hardware in over 10
years. Meanwhile, mainstream computers have increased in speed by a
couple of orders of magnitude.
>
> One story I've heard is that back in the 1980s, American Express
> developed an expert system (which might still be in use) for approving
> credit card transactions. As I understand, AmEx doesn't assign a
> specific credit limit to an account; rather, they approve transactions
> on a case-by-case basis depending on various factors such as the
> account's existing balance, payment history, size of the requested
> transaction, and sizes of past transactions. (So e.g. if you have a
> history of making large purchases and paying them off promptly,
> they'll keep allowing you to do that.)
>
> Anyway, I recall being told that somewhere AmEx had a bank of
> Symbolics machines running this expert system, making these
> transaction approval decisions in real time. That may not be exactly
> what you have in mind, but I think it demonstrates that these machines
> were perfectly capable of providing E-commerce services.
AmEx was also a Thinking Machines customer, and I think much of the data
mining in that expert system was done on the Connection Machine, not the
Lisp Machine. But the Lispm was the front-end controlling the CM, IIRC.
--
Barry Margolin, ······@alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***
*** PLEASE don't copy me on replies, I'll read them in the group ***
In article <····························@newsgroups.comcast.net>,
Barry Margolin <······@alum.mit.edu> wrote:
> In article <·······················@i38g2000prf.googlegroups.com>,
> Scott Burson <········@gmail.com> wrote:
>
> > On Jul 29, 2:54 pm, gavino <·········@gmail.com> wrote:
> >
> > > has anyone ever used a lisp machine as an e commerce server? (aka not
> > > just a workstation)
>
> I don't think there has been any new Lisp Machine hardware in over 10
> years. Meanwhile, mainstream computers have increased in speed by a
> couple of orders of magnitude.
>
> >
> > One story I've heard is that back in the 1980s, American Express
> > developed an expert system (which might still be in use) for approving
> > credit card transactions. As I understand, AmEx doesn't assign a
> > specific credit limit to an account; rather, they approve transactions
> > on a case-by-case basis depending on various factors such as the
> > account's existing balance, payment history, size of the requested
> > transaction, and sizes of past transactions. (So e.g. if you have a
> > history of making large purchases and paying them off promptly,
> > they'll keep allowing you to do that.)
> >
> > Anyway, I recall being told that somewhere AmEx had a bank of
> > Symbolics machines running this expert system, making these
> > transaction approval decisions in real time. That may not be exactly
> > what you have in mind, but I think it demonstrates that these machines
> > were perfectly capable of providing E-commerce services.
>
> AmEx was also a Thinking Machines customer, and I think much of the data
> mining in that expert system was done on the Connection Machine, not the
> Lisp Machine. But the Lispm was the front-end controlling the CM, IIRC.
The reports about this application never mentioned
Thinking Machines. Do you have more details?
The application itself is a rule-based
system developed with ART (an expert system tool). A report from
2002 says, it is still in use, running now on AIX machines.
http://bizrules.info/page/art_amexaa.htm
http://www.ddj.com/184405217
--
http://lispm.dyndns.org
From: John Thingstad
Subject: Re: I symbolics still selling machines? does anyone use the lisp genera os?
Date:
Message-ID: <op.tv8qc31epqzri1@pandora.upc.no>
P� Sun, 29 Jul 2007 14:08:25 +0200, skrev Frank Buss <··@frank-buss.de>:
> Then you can measure the number of lines:
>
> find . -exec cat {} \; | wc
> 8,499,410 29,654,309 244,866,374 (lines, words, chars)
>
> find . -type f -name *.c -exec cat {} \; | wc
> 5,979,935 19,411,234 16,2966,504
>
> find . -type f -name *.h -exec cat {} \; | wc
> 1,462,784 6,084,899 52,288,919
>
This might be nit picking but..
find . -type f \( -name '*.c' -or -name '*.h' \) -exec cat {} \; | wc -l
is a bit more succinct (the -l option just gives the lines)
John Thingstad wrote:
> This might be nit picking but..
>
> find . -type f \( -name '*.c' -or -name '*.h' \) -exec cat {} \; | wc -l
>
> is a bit more succinct (the -l option just gives the lines)
yes, that's better, if you want to sum C and H files, but if you sum them
separately may give you more information.
Interesting would be the number of code lines as defined in the referenced
article ( http://www.dwheeler.com/sloc/redhat71-v1/redhat71sloc.html )
without comments and empty lines and without duplicate files, but for
kernel 2.6. This is the tool: http://www.dwheeler.com/sloccount/
Has anyone the Lisp machine source to run this tool on it?
--
Frank Buss, ··@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de
In article <·······························@40tude.net>,
Frank Buss <··@frank-buss.de> wrote:
> John Thingstad wrote:
>
> > This might be nit picking but..
> >
> > find . -type f \( -name '*.c' -or -name '*.h' \) -exec cat {} \; | wc -l
> >
> > is a bit more succinct (the -l option just gives the lines)
>
> yes, that's better, if you want to sum C and H files, but if you sum them
> separately may give you more information.
>
> Interesting would be the number of code lines as defined in the referenced
> article ( http://www.dwheeler.com/sloc/redhat71-v1/redhat71sloc.html )
> without comments and empty lines and without duplicate files, but for
> kernel 2.6. This is the tool: http://www.dwheeler.com/sloccount/
> Has anyone the Lisp machine source to run this tool on it?
RJMBP:/Lisp/software/genera-8-3/sys.sct joswig$ find . -type f \( -name '*.lisp*' \) -exec cat {} \; | wc -l
1360963
Which is much (not all) of Genera plus some other bits...
--
http://lispm.dyndns.org
On Jul 29, 2:08 pm, Frank Buss <····@frank-buss.de> wrote:
> ·········@gmail.com wrote:
> > Is the Linux's 2.4 million SLOC just the size of kernel? Or includes
> > device drive?
>
> It doesn't make sense to compare this. Maybe there was some useful software
> in the Symbolics code, but I doubt that it had drivers for all kind of
> hardware like Linux, e.g. Linux supports more than 60 network cards (http://tinyurl.com/34zclc) and I assume there are lots of device types the
> Symbolics machine not even know, but the Lisp machine has some application
> code, which the Linux kernel doesn't have.
>
> You can download the Linux kernel fromhttp://www.kernel.org/, e.g. the
> latest kernel:ftp://ftp.kernel.org/pub/linux/kernel/v2.6/linux-2.6.22.tar.gz
> Then you can measure the number of lines:
>
> find . -exec cat {} \; | wc
> 8,499,410 29,654,309 244,866,374 (lines, words, chars)
>
> find . -type f -name *.c -exec cat {} \; | wc
> 5,979,935 19,411,234 16,2966,504
>
> find . -type f -name *.h -exec cat {} \; | wc
> 1,462,784 6,084,899 52,288,919
>
> I don't know which kernel version the blogger used for his 2,4 million
> lines of code. Adding the headers and C files, the Linux kernel has 7,4
> million lines of code.
>
> Symbolics Lisp machines were sold with the computer, so maybe it supports
> one network card? This means it would need 460,000 * 60 = 28 million lines
> of code to support all drivers Linux does. The conclusion: C is nearly 4
> times more efficient in code size than Lisp :-)
>
> --
> Frank Buss, ····@frank-buss.dehttp://www.frank-buss.de,http://www.it4-systems.de
Please excuse my hardware ignorance but aren't device drivers
implemented separately from the kernel?
Slobodan Blazeski wrote:
> Please excuse my hardware ignorance but aren't device drivers
> implemented separately from the kernel?
You are right, but the package which is called "the Linux kernel source" at
http://www.kernel.org/ includes the core kernel and all drivers, which can
be compiled into the kernel (good for embedded systems) or as loadable
modules.
--
Frank Buss, ··@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de
On Jul 29, 6:31 am, gavino <·········@gmail.com> wrote:
> http://bc.tech.coop/blog/040128.htmlwas reading here about linux
> kernel lines vs common lisp lines of genera...wow!! 4e5 vs 2e6 lines
To answer the question in your subject line, which seems to be
unrelated to the text, yes, Symbolics is still selling Lisp machine
hardware, as well as the Genera OS to run on AlphaStations. Contact
·····@symbolics.com for more information.
Disclosure: I just bought a MacIvory 2 from them last week, as
mentioned on my blog http://jaoswald.wordpress.com/2007/07/11/ordering-a-lisp-machine/