Surely it's at higher levels that the compiler
has enough context to make better optimizations.
If you present an assembler with a program there
is very little that can be done to optimize it
without knowing exactly what the programmer
intended (which is computationally infeasible),
whereas the semantically more meaningful constructs
in a HLL make it easier for the compiler to reason
about the program.
The result of the real player case was that it turned out it was a bug in their software as I remember. I don't know aobut Lotus 123 but when service pack 6 broke Lotus Notes MS had to release a new version very quickly to fix it.
> 2) Microsoft has the ability to change its > API at will. > Yes and no. Microsoft can add new APIs but they are constrained by the number of applications available as far as changing old ones goes.
Internet Explorer does work with regular HTML. It may not implement all of the standards but then nothing in mainstream use does (mozilla certainly doesn't). But in addition to standard HTML IE also implements some additional features. So what! No one is forcing website designers to use them, any more than they were when Netscape introduced the BLINK tag and frames.
Well a company can own the market for a copyrighted material like a particular drug. In the case of uncopyrightable material like software, they can raise the barriers to entry by keeping their sourcecode and API definition as trade secrets. If you publish APIs then competitors can enter the market with identical products and forego the marketing cost involving in differentiating the original product from similar but non-identical competitors. If you publish sourcecode (even under the GPL) then there are really no barriers to entry at all. Of course we in the large community might view this as a good thing because of the effect on prices. Alternatively it might be a bad thing because of the disincentive to invest in initial development of new software.
> If a potential customer of yours sees more > value in reimplementing your stuff than buying > it from you > I would have thought the problem would be with competitors free riding on the expense involving in creating a market for implementing these particular APIs.
I can't really see why the improvements they suggest justify a new language. Garbage collection, automatic initialization and inbuilt support for COM are features that could easily be added to C/C++ (and have been). Perhaps the real intention is to replace VB (a nice idea IMHO but really suffering from poor design and the a tendency to add new languages element willy-nilly) with C#.
Well I've add the power supply fail on my NT box a couple of times and it takes very little extra tiime to start up if the filesystem wasn't cleanly unmounted, certainly at lot less time than it takes to fsck a similar sized drive under linux. NT definitely has a working log filesystem (not quite the same as a transaction one AFAIK) but it may have other bugs. Perhaps you should call your helpdesk instead of bitching on slashdot:-)
Yeah, it seems from the discussions on linux-kernel that having a transaction filesystem leading to problems with too many pages being dirty because the vm system can't write out ones that are required for a transaction, so some sort of major modification is required to fix things. NT has a similar write-throttling mechanism.
> checkout out tpc.org > I'm not sure this is quite true (though Win2K is a great operating system), wasn't the benchmark used for the Win2K tests one that was completely parallelizable (sp?) so a group of servers could run different bits independantly without the usual overheads of distributed solutions.
> it's perfectly legal for me to create a website > with naked 16 year old school girls on it > Wouldn't this fall under existing British obscenity laws?
I find gdb crashes frequently when trying to debug any non-trivial threaded program. Mozilla have a whole page of complaints about it http://www.mozilla.org/unix/debugging-faq.html.
No, the reason microsoft implemented these extensions was to carry information about what groups and permissions a user has. The reason the field they used was added to the Kerberos protocol was for just such a purpose. What was objectionable IMHO was making the details of their extensions a trade secret.
Only if linking with a library causes the whole program (as a work derived from the library and the GPLed code) to have additional restrictions that are incompatible with the GPL. It is quite possible to run GPL programs under non-GPLed operating systems, as all the FSF did before Linux came along.
Only if it runs in user-mode. There maybe additional issues with a kernel-mode NFS server.
Re:This all prooves MS is good for consumers!!
on
Gnutella VBS Worm
·
· Score: 1
If you want to exploit bugs in MS operating systems you will still need a good grasp of assembler (or least to be able to get an exploit from someone who has). If you want to write a malicious program you could write it in VBS, C, FORTRAN for any kind of computer. Trojan horse programs don't require any special knowledge other than a bit of social engineering and never have.
Basic communication with the hardware is done by reading/writing i/o ports. The advantage of the BIOS for DOS was that it hid the details of the different types of hardware and provided a uniform interface. Modern operating systems like UNIX or NT use device specific drivers that are specifically written for them to perform a similar function.
Disk access under Linux bypasses the BIOS so it shouldn't be a problem. Though virus checkers run under Windows might report a problem if the boot sector has changed.
Slashdot Mirror
Surely complete optimization is equivalent to
the halting problem?
Surely it's at higher levels that the compiler
has enough context to make better optimizations.
If you present an assembler with a program there
is very little that can be done to optimize it
without knowing exactly what the programmer
intended (which is computationally infeasible),
whereas the semantically more meaningful constructs
in a HLL make it easier for the compiler to reason
about the program.
The result of the real player case was that it
turned out it was a bug in their software as I
remember. I don't know aobut Lotus 123 but when
service pack 6 broke Lotus Notes MS had to release
a new version very quickly to fix it.
> 2) Microsoft has the ability to change its
> API at will.
>
Yes and no. Microsoft can add new APIs but they
are constrained by the number of applications
available as far as changing old ones goes.
Wouldn't entering into negotiations have been taken
by the WIPO as evidence he was a cyber squatter?
I'm not sure which unices you're referring too,
but linux threads suck. Debugging any non-trival
threaded program is a nightmare.
LOL!
go_to_the_park is_it_raining = if is_it_raining then "Yes" else "No"
(go_to_the_park in Haskell)
Dude,
Internet Explorer does work with regular HTML.
It may not implement all of the standards but
then nothing in mainstream use does (mozilla
certainly doesn't). But in addition to standard
HTML IE also implements some additional features.
So what! No one is forcing website designers to
use them, any more than they were when Netscape
introduced the BLINK tag and frames.
Well a company can own the market for a copyrighted material like a particular drug. In the case of uncopyrightable material like software, they can raise the barriers to entry by
keeping their sourcecode and API definition as
trade secrets. If you publish APIs then competitors can enter the market with identical
products and forego the marketing cost involving
in differentiating the original product from
similar but non-identical competitors. If you
publish sourcecode (even under the GPL) then
there are really no barriers to entry at all.
Of course we in the large community might view
this as a good thing because of the effect on
prices. Alternatively it might be a bad thing
because of the disincentive to invest in initial
development of new software.
> If a potential customer of yours sees more
> value in reimplementing your stuff than buying
> it from you
>
I would have thought the problem would be with
competitors free riding on the expense involving
in creating a market for implementing these
particular APIs.
I can't really see why the improvements they suggest
justify a new language. Garbage collection,
automatic initialization and inbuilt support for
COM are features that could easily be added to
C/C++ (and have been). Perhaps the real intention
is to replace VB (a nice idea IMHO but really
suffering from poor design and the a tendency
to add new languages element willy-nilly) with C#.
Well I've add the power supply fail on my NT :-)
box a couple of times and it takes very little
extra tiime to start up if the filesystem wasn't
cleanly unmounted, certainly at lot less time
than it takes to fsck a similar sized drive
under linux. NT definitely has a working log filesystem
(not quite the same as a transaction one AFAIK) but it
may have other bugs. Perhaps you should call
your helpdesk instead of bitching on slashdot
Yeah, it seems from the discussions on linux-kernel that having a transaction filesystem
leading to problems with too many pages being
dirty because the vm system can't write out ones
that are required for a transaction, so some sort
of major modification is required to fix things.
NT has a similar write-throttling mechanism.
> checkout out tpc.org
>
I'm not sure this is quite true (though Win2K is
a great operating system), wasn't the benchmark
used for the Win2K tests one that was completely
parallelizable (sp?) so a group of servers
could run different bits independantly without
the usual overheads of distributed solutions.
> it's perfectly legal for me to create a website
> with naked 16 year old school girls on it
>
Wouldn't this fall under existing British
obscenity laws?
I find gdb crashes frequently when trying to
debug any non-trivial threaded program. Mozilla
have a whole page of complaints about it
http://www.mozilla.org/unix/debugging-faq.html.
No, the reason microsoft implemented these
extensions was to carry information about what
groups and permissions a user has. The
reason the field they used was added to the
Kerberos protocol was for just such a purpose.
What was objectionable IMHO was making the
details of their extensions a trade secret.
www.openbsd.org is actually
openbsd.sunsite.ualberta.ca, www.usa.openbsd.org
is actually running OpenBSD itself.
Only if linking with a library causes the whole
program (as a work derived from the library and
the GPLed code) to have additional restrictions
that are incompatible with the GPL. It is quite
possible to run GPL programs under non-GPLed
operating systems, as all the FSF did before
Linux came along.
Only if it runs in user-mode. There maybe
additional issues with a kernel-mode NFS server.
If you want to exploit bugs in MS operating
systems you will still need a good grasp
of assembler (or least to be able to get an
exploit from someone who has). If you want
to write a malicious program you could write it
in VBS, C, FORTRAN for any kind of computer.
Trojan horse programs don't require any special
knowledge other than a bit of social engineering
and never have.
If a language allows side-effects it isn't
a *pure* functional language. Even languages
like Haskell have limited side-effects to implement
I/O.
You can use other languages in ASP pages, JavaScript, Perl even Haskell.
Basic communication with the hardware is done
by reading/writing i/o ports. The advantage of
the BIOS for DOS was that it hid the details of
the different types of hardware and provided a
uniform interface. Modern operating systems like
UNIX or NT use device specific drivers that are
specifically written for them to perform a similar
function.
Disk access under Linux bypasses the BIOS so
it shouldn't be a problem. Though virus checkers
run under Windows might report a problem if the
boot sector has changed.