The encoding is not the real problem. Also, the send-it-everywhere approach to distributing articles isn't a big problem for binaries. It's efficient enough in real life, though it does limit what can be done in the more interactive [text] parts of Usenet. Binary groups are't really intereactive (well, fnarr I guess, but...).
The use of a protocol and article format designed for text messages rto distribute stuff that isn't text is the problem.
There are ready-made solutions, even standardized ones, to all of this. It's a (not necessarily simple, but you might be surprised) matter of reaching into the Obvious Bag, pulling out the stuff that's already made to do the job, and coding it.
GNU carries 18 years of its own emotional baggage, and it was intended from the beginning to be a superset of Unix. This isn't the clean slate you're looking for.
Cocoa was already given a shot on the intel platform in its previous incarnation from NeXT. This didn't create incredibly huge sales figures before the merger with Apple, so why would it now?
Basically, when you bought the product, you would receive the source, with a license stating that you could only share the source and any modifications with other license holders of the software, as well as with the company - in fact, make it a necessity that the company gets the mods , so the new users/buyers of the software get the mods as well in their copy of the source
This is essentially the license Lucent use to make Plan 9 available, except that they no longer charge for the software.
If you're got a POP3 swerver running, does it support XTND XMIT? Not everyone can use that, but Eudora can send mail that way and *x users can use a little script to replace sendmail.
It's little and I'd post it, but the lameness filter is too lame to let it though. Bah.
Well... perl -T doesn't offer any sort of guarantees. It's not perfect (as bug fixes to taint mode over time will attest) and can (intentionally or accidentally) be bypassed using the documented laundering techniques. That's not to say that the taint mechanism is not useful -- it is a help -- but one must take care not to be lulled into a false sense of security in its capabilities.
With all those qualifiers out of the way: yes, Python does try to solve simliar problems with its bundled rexec and Bastion modules. Again, this should not be seen as a magic bullet! With enough effort, people can accomplish stupid things in any environment.
SPKI/SDSI is quite a the nifty way to deal with some notoriously hairy auth problems that traditional hierarchical PKI never really satisfied. Unfortunately, people seem to have difficulty in understanding it, and that in itself is a problem. In some ways it doesn't much matter that a system is adequately secure if its users can't see why they should be able to trust it. Maybe it does just take implementing the thing and letting time take its course, but that's not very much fun.
Slashdot Mirror
The encoding is not the real problem. Also, the send-it-everywhere approach to distributing articles isn't a big problem for binaries. It's efficient enough in real life, though it does limit what can be done in the more interactive [text] parts of Usenet. Binary groups are't really intereactive (well, fnarr I guess, but...).
The use of a protocol and article format designed for text messages rto distribute stuff that isn't text is the problem.
There are ready-made solutions, even standardized ones, to all of this. It's a (not necessarily simple, but you might be surprised) matter of reaching into the Obvious Bag, pulling out the stuff that's already made to do the job, and coding it.
Hm, this story again.
Aren't these guys still making their own tapes?
GNU carries 18 years of its own emotional baggage, and it was intended from the beginning to be a superset of Unix. This isn't the clean slate you're looking for.
Cocoa was already given a shot on the intel platform in its previous incarnation from NeXT. This didn't create incredibly huge sales figures before the merger with Apple, so why would it now?
The problem for Google may be to decide upon the definitive article.
Would it be best to mark this incident with big words and perfect grammar, the initial declaration of Meow, or the emergence of Fluffy?
Suggestions?
If you're got a POP3 swerver running, does it support XTND XMIT? Not everyone can use that, but Eudora can send mail that way and *x users can use a little script to replace sendmail.
It's little and I'd post it, but the lameness filter is too lame to let it though. Bah.
Well... perl -T doesn't offer any sort of guarantees. It's not perfect (as bug fixes to taint mode over time will attest) and can (intentionally or accidentally) be bypassed using the documented laundering techniques. That's not to say that the taint mechanism is not useful -- it is a help -- but one must take care not to be lulled into a false sense of security in its capabilities.
With all those qualifiers out of the way: yes, Python does try to solve simliar problems with its bundled rexec and Bastion modules. Again, this should not be seen as a magic bullet! With enough effort, people can accomplish stupid things in any environment.
SPKI/SDSI is quite a the nifty way to deal with some notoriously hairy auth problems that traditional hierarchical PKI never really satisfied. Unfortunately, people seem to have difficulty in understanding it, and that in itself is a problem. In some ways it doesn't much matter that a system is adequately secure if its users can't see why they should be able to trust it. Maybe it does just take implementing the thing and letting time take its course, but that's not very much fun.