While what happened in your class was pretty clearly cheating (sale of results by one student), I'm bothered by similar stories I hear in which several students collaborate on methods, and tend to converge on similar (though usually not identical) code.
IMO, this second case is a valuable lesson in the power of group development, sharing ideas, and the open source methods described by Eric Raymond in The Cathedral and the Bazaar. Much more valuable than the typical function-and-methods lessons taught in CS programs. Education has the problem of trying to assign credit for work performed, but an enlightened instructor could probably work out some compromise method.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
Jamie specifically noted that only one of several developers responded to his queries:
Getting any developers to talk about this bug has been like pulling teeth. Only one of the developers I contacted (repeatedly) even bothered to return my email.
(Maybe they've never heard of Slashdot?)
This is consistant with several other reports -- I'd also exchanged email with Blake Ross. Two emails (one asking if he'd posted to Kuro5hin) netted a grand total of "that's me:)" in response.
While I haven't seen compelling evidence of skulduggery in this case, Netscape/AOL have been less than forthcoming.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
Actually, I'd prefer a read-only source license. Reveals secrets without facilitating open source development.
Of the options you've given, old-style BSD. I'd like to encourage forking, including proprietary forking, of MSFT OS and apps. I'd also like it to be impossible to convert or merge it to GPL. No copylefting here.
Yes, this is punitive in intent.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
...and what's to keep you from upping the ante on the computational difficulty. Say your problem is "brute force this hash". Increasing the keylength by one byte doubles the (mean) compute time. Make this user configurable, or if you're really smart, code the program to generate problems which require some specified mean time to compute over recent requests recieved, or such.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
Unfortunately, the USPS's change of address form feeds directly into a database which is, you guessed it, sold to marketing firms.
The turnaround time for propogating the information out is about three to six months, so for the first half of a year, you're ok, but afterwards, you've become identified as a high-quality (recent) address, and to boot, recently moved so a good candidate for household goods, services,....
What part of "Gestalt" don't you understand? Scope out Kuro5hin
I noted that the Minix license, as quoted, does not include the advertising clause. What part of this do you not understand?
I referred to the change notice posted by UC Berkeley because I couldn't find a copy of the full text of the earlier version of the BSD license. However, it is the same clause 3 which effects the advertising clause. You'll note (if you bother to follow the two thoughtfully provided links) that this clause three isn't present in the Minix license.
Minix was never issued under the prior version of the BSD license, so no, the change notice doesn't affect Minix. You appear to be confused on this point.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
While having a useful teaching tool is a laudable goal, there's no reason why the project couldn't fork, with one version of the kernel being a simple, understandable kernel, and the other serving as the basis for more complex (and presumably useful) designs.
Both the BSD and GPL licenses make forking possible. What the GPL does is ensure that, to the extent anyone wishes to do so, there are no legal embarrments to merging forks down the road. If a full-featured Minix (Maxix?) were to implement some trick-cool concept, that feature and it alone, could be merged back into Minix.
While fork/remerge is possible with BSD code which remains BSD licensed, proprietized development based on BSD code cannot be remerged back to the original core without explicit permission of the new copyright holder.
IMO Tanenbaum's probably too little, too late, though it would be interesting to see what shows up under system software at SourceForge over the next few months....
What part of "Gestalt" don't you understand? Scope out Kuro5hin
The license is quoted in full at the Deja archive. The advertising clause afflicted only the "old" BSD license, the language is included in the notice which rescinded this clause.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
Ulterior motives are fine, so long as there's a balance of them among those who hold power. It's when everyone's got the same ulterior motive that I start worrying.
My ulterior motives are derived from free software. You gotta problem w'dat?
What part of "Gestalt" don't you understand? Scope out Kuro5hin
Dude, what are you doing posting meaningful stuff on Slashdot for?
Yes, a university which respected its student body would refer a situation such as this to its academic judiciary. It seems from the facts presented that this isn't the case here -- the school doesn't respect its students enough to supply what's rapidly becoming base-level essentials. They're going ballistic for creative appropriation of what is already freely available. Excellent suggestion, snowball's chance in hell.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
NT qualifies under UNIX98 branding as a Unix system when running the compatibility overlay previously known as Interix/OpenNT. Covered in this ZDNet article. MVS has also qualified for Unix branding, IIRC.
Not that I think of NT as Unix.
What part of "Gestalt" don't you understand? Scope out Kuro5hin
Cookies are broken. They've outlived their usefulness, and are hopelessly open for abuse.
I have two suggestions:
For single-session state tracking, cookies serve a purpose -- in fact this is largely what they were designed for. To this end, allowing cookies -- for the duration of a single browser session and possibly less -- may be a legitimate use.
For authentication and account-state tracking, stronger, more user-controlled, and less spoofable means are required. One technology already exists -- public key encryption and challenge-response based authentication.
The first suggestion would allow cookies to be used to track navigation and state through a single session at a site. The functionality is already available in a browser such as Netscape Navigator if you link your cookie file to/dev/null (Linux/Unix) or to a directory (Windows). Cookies are accepted but not permanently stored on your system. The upside is that cookie-dependant features of sites work. The downside is that state such as user ID and passwords have to be re-entered for each browser session.
PKE/CRA would work based on public/private key pairs, as with PGP. A user could generate as many or few of these key pairs, and optionally share them (both public and private) with other users, as desired. On entering a site requiring registration, the user could choose the key (the session identity) to send the site. If a private, secret identity is chosen, the session is personal. If a generally known key (say, cypherpunks) is sent, the session is authenticated, but not private. The remainder of the session is transacted over secure links (SSL), and cookie or other state-tracking could be used to register and/or log activity.
The strength of this scheme is allowing a user to specify both the degree of authentication, and identity authenticated used when browsing sites. If desired, keys could be generated and destroyed on a regular basis, reducing the utility of any tracking of keys. Control over whether to authenticate, who to authenticate to, and who to authenticate as, is left to the user.
Existing browser technology has been driven very strongly by server-side interestes -- user tracking, profiling, and e-commerce vendor desires. The interests of the user have not been represented, and are only partially filled by such patches as IDcide and Junkbuster (I'm another satisfied JB user). We've got the source, and with it the ability to reclaim the power.
Interesting observation. I guess in a way SAS is something of a Perl equivalent for the mainframe -- general, high-level, all-purpose, interpreted tool. Comparatively it's relatively inexpensive (though by no means cheap), as well.
There are other MVS scripting tools though -- JCL has certain capabilities. REXX is also there, though I believe it's a relative newcomer. SAS's advantage is that it eats many common MVS file formats with relative ease. The same file-based inter-process communication method is used in both SAS programs and MVS jobs. Given that many MVS processes are heavily task, batch, and data oriented, it's a reasonably good fit.
The main differences on Unix are that there is much more system interactivity -- jobs can be interactive or trigger interactive events -- and interprocess communication often occurs as a data pipe rather than a disk file. A data step or SQL view approximates the latter. Pipelines save scads of disk I/O, and are a really cool feature. If an entire set of processes can be pipelined (data streamed from beginning to end), processing can be much faster than if a file or dataset has to be generated and read at each stage.
Unix is script-tool rich -- shells, sed, awk, Perl, Python, Tk,.... Yes. While some scripting tasks could be handled by SAS, you'd in general try to avoid doing this.
SAS's announcement is really great news for Linux, and it's something that users had been requesting for years, going back to a request by Christa Keil in 1993. There is no single solution providing the depth and breadth of data analysis and processing capabilities as SAS. I spent seven years as a SAS developer for clients in healthcare, banking, policy research, finance, biotech, and pharmaceutical sectors.
Two years ago I launched the SAS4Linux website and mailing list to focus the campaign. While it did generate interest from users, and responses from SAS Institute, I tend to think SI arrived at its own conclusions in deciding to market the port.
As others have noted, there is a bit of "too little, too late" from a number of folks as well. I now count myself as a former (recovering?) SAS programmer. I've been sufficiently infected with free software concepts to feel that there are several things about SAS which could be improved. In fact, many of them have been, though they are approached in a number of different, and distinct, open source projects.
The obvious next step is to provide some duct tape and wrapping around these tools to provide similar functionality (though not run-alike compatibility) to SAS. Stay tuned;-)
As I read the announcement, the port was for "the SAS System". Presumably this includes all system components.
What I understand of the SAS architecture is that it is largely platform independent, with the dependent stuff largely isolated into a (small?) number of modules. Probably less than 10% of the codebase. Meaning that you get a lot of portability bang for your programming buck.
The SAS C Compiler is what SI themselves developed to build SAS itself. So it was probably one of the earlier technical ports to the Linux platform. Moreover, it's been generally known for a couple of years that SI have been running a compile farm based on FreeBSD, particularly for their Mainframe products. Smoke that.
My guess is that SAS/Toolkit (the developers library) will be availble for Linux -- at a price. C'est la vie.
Which then raises the obvious question: are the 1201 anti-circumvention protections copyright-like protection of expression, or patent-like protection of ideas, and if the latter, how are they affected by decisions in MAI v. Peak and Sony v. Connectix?
The irony is biting, particularly as I just went through a round of emails with Rob last week regarding off-topic meta-commentary on Slashdot in various posts. I'd run across an announcement of the Slashdot/VA merger in an unrelated article -- prior to Slashdot's posting notice of the announcement -- which had been moderated "Offtopic". I contested the rating, and the general practice of down-moderating Slashdot-topic posts, see below for salient commentary.
Date: Wed, 16 Feb 2000 08:32:55 -0500 (EST) From: Rob Malda <malda@slashdot.org> To: kmself@ix.netcom.com cc: cmdrtaco@slashdot.org Subject: Re: Metamoderation comments
On Tue, 15 Feb 2000 kmself@ix.netcom.com wrote:
> While the moderation is technically true, the fact is that there is *no* > place on Slashdot to address news or issues concerning Slashdot itself. > While the post is off _topic_, I feel it is on _meta_topic, and that > the moderation was inappropriate.
I think its completely offtopic and deserves to be moderated down.
People need to chill out and wait until a story comes along *about* Slashdot and talk *there* about it. At least then (surprise) I might read it and do something about it.
-- | Rob "CmdrTaco" Malda | Just want to be misunderstood, | malda(@)slashdot.org | I wanna be feared in my neighborhood. | http://slashdot.org/ | Just want to be a moody man, | cartoonist*lovemachi | say things that nobody can understand. | ne*obsessivecompulsi | --Pete Townshend, Misunderstood
My point: Slashdot metacommentary is always, if not on topic, at least on metatopic. What's particularly ironic in this case is that Rob could have created an article to post his rant to, and ensuing commentary, but instead chose to post it here, despite what he'd said to me last week.
It's not the off-topicality that I'm objecting to in Rob's rant, it's the hypocrisy.
And I continue to request that there be a persistant, front-page linked, Slashdot-on-Slashdot forum where these conversations are on topic. Failing to provide such a forum, while encouraging the marking of such posts off-topic, is a subtle form of censorship.
Mmmmph. I don't know about this. Is it substantially different from using a +1 score threshold if you don't want to see the "unvetted" AC comments? I suppose one might miss less with your solution, but instructing moderators to focus on raising good AC posts above 0 and increasing the number of points + moderators available might alleviate that.
Possibly, depending on the implementation.
If AC posts are relegated to "purgatory" rather than the main message board, and access to purgatory were limited to registered users, registered users of sufficient karma, or active moderators, then the positive reinforcement cycle of trolling (look at what I did to the board) is lost. ACs can't see their own trolls, and the larger public doesn't see them either. Not that the AC boards can't be spammed, but the psychological reward of doing so is greatly reduced.
Technical protections (post lockouts, redundancy filters, etc.) may still be required, but the main board problems should be much reduced.
Also agreed. Along those same lines, and somewhat aside, I'd like to see stories be assigned more than one topic (eg a story about SGI supporting Linux doesn't just get in the "Linux" category or just the "SGI" category.) Along with which users should be granted more flexibility in filtering what stories they see on the main page.
That's getting toward keywording rather than departmentalizing stories. IMO it's a good idea, and becomes necessary when either traffic or classification schemes become sufficiently rich. Face it, the world ain't heierarchical.
There are two reasons I disagree with the "off topic" classification:
In this and the prior article, spammage and brokenness of the moderation system are directly reflected in the article follow-ups.
Slashdot offers no sanctioned "meta article" for discussiong Slashdot community issues, particularly moderation. This is an issue I've raised with Rob and Jeff several times. The sid=moderation underground forum is an option, but its existence isn't widely known.
I have a serious problem with Slashdot community threads being marked off-topic, as they generally emerge in response to a blatantly evident problem, and no alternatives exist. Moderating these threads down shows an overly literal reading of the moderation guidelines, IMO.
There would be a link between a user and an anonymous post in the database, which would allow that information to be subpeonaed into court, should it be required. That defeats entirely the whole philosophical point behind anonymous postings.
Sorry, I don't follow.
There is no link between the identity of the AC posting, and the comments posted by the AC. There is a link between the Slashdot ID of the person signing the post, and the post, but this implies absolutely nothing WRT origin of content.
What specific legal threats do you see as originating from the signing of anonymous posts to allow general viewing? AFAIK, there is no transitivity of libel from the person writing it to a person who allows it to be generally posted. The original poster's anonymity is preserved.
There is a "hidden" forum at sid=moderation on the topic of moderation. While useful, comments posted to it have the unfortunate tendency to disappear after time.
There are obviously several very broken things about/. which are defeating the moderation system. These include troll posts, the apparent inability to restrict nonsensical or repeated posts, and an serious shortage of moderation points to posts. With karma 44, I've had moderation privileges once this millenium -- prior to December, 1999, it seemed I had mod priv typically once or twice a week. Comments get caught in a vicious cycle of failing to be moderated up once an article topic has reached a hundred posts or more -- for meaningful use, I'm forced to read at score=1, score=2, or higher.
Repeating comments I've made (and have since slipped from) sid=moderation:
There need to be far more moderation points. The ratio of points to posts needs to increase at a rate greater than traffic.
Anonymous coward posting needs to be seriously rethought. CmdrTaco will insist on it being allowed. I'd propose a "Yale Wall" solution (See Lawrence Lessig's Code and Other Laws of Cyberspace, p79.
Essentially, the idea is that AC posts would require vetting by at least one registered user. The user would face the karmic consequences of any moderation of the post, and would grant the AC post any karmic benefits: If the post is moderated up or down, the signer's karma is increased or reduced, and whatever default posting level the signer has is given to the post. This preserves anonymity while allowing a modicum of control over the AC process.
To do this, there would have to be a seperate viewing field for unvetted AC posts. Once vetted, the posts would enter the main forum and be subject to moderation. Only one vetting would be required to transfer the post to the main board.
The existing karma/default score system is working fairly well. The main problems are flagrant abuse by a very small number of individuals (whoring points, then spamming), and AC posts. The first problem could be addressed by administrative karmic reduction for spammage. The AC problem would go away with the vetting system, a default score of "unmoderated", and ranked moderation (see below).
A distinction must be made between posts which have and have not yet been moderated, and a moderation filter must exist to allow viewing (and moderating) of previously unmoderated posts. It's simply too painful to prowel through posts looking for that which should be moderated, up or down. The best current option is to view newest first w/o threads. It's still a poor proxy.
Moderating and posting to a forum should be allowed. CmdrTaco doesn't like it. Fooey. Moderating your own posts should not be allowed. Abuse can be detected in M2.
Notification of current moderation status of a post should be given before moderation points are committed. It's often the case when moderating an active forum that several people have moderated the same post(s) simultaneously. A confirmation screen saying "this post is currently moderated at [score], continue?" should be given. Wasting points shooting down crap is plain stupid. Boosting a decent but not brilliant post to +5 is silly.
Better filtering options. Score based filters are limited. It should at least be possible to set a ceiling as well as a floor (show all posts between -1 and 1). Better yet would be to allow filtering out posts according to criterion. I'm annoyed frequently by amusing, but otherwise not very interesting posts, which are highly scored. They're often the highest scored in an article.
I'd like to see moderation transformed from a voting system to a rating system. Essentially, now, a moderator can score a post up or down, and cumulative score is what counts. IMO, taking a weighted average of point scores -- say -5 to +5 -- for a post, would be a useful system. The number of points which could be voted, both on a single post and within a single moderating session, could be tied to karma. This would allow a moderator to express mild to high approval or disapproval of a post. Some measure of interest or intensity could be derived from the number of ratings, as well as the aggregate (mean) rating, and an inference of controversiality from the standard deviation (for all you stats hounds out there).
There are other problems. Karma whoring, gullibility traps (posts written to look informative but actually false), etc. I believe that as other issues in the moderating system are dealt with, the magnitude of these issues will be diminished. Or they won't <g>. But there are bigger fish to fry first.
The idea of an RBL type system is something I've thought of independently. It seems attractive. Like the UDP and real RBL, it could be a loose affiliation, decentralized, and advisory in nature. No need to bring the government in -- little that it could likely do anyway.
Realistically, what would be required is for a given network gateway to monitor its peer and child connections. Portscanning might not be necessary, depending on the signatures of an attack. A particular peer/child which exhibited behavior indicative of compromised host(s) could be blocked off, with appropriate messages sent to administrative contacts.
At the ISP level, this would include monitoring both individual dialup/fixed IP hosts, and connections to other IP aggregators. A sufficient level of filtering/blocking would act like a circuit breaker -- portions of the net might be slowed or cut off, but global abuses of the sort experienced in the past few weeks would be avoided.
One thing I haven't seen in any news stories or most of the commentaries posted is what specific hosts and operating systems are being compromised. There was the withdrawn story to Computer Currents yesterday which claimed only Linux and Solaris were involved. I find this hard to believe. I've heard anectdotal evidence that Windows machines are the most frequently compromised hosts, via viruses.
If the truth is lurking somewhere in earshot, could it please make itself heard?
Slashdot Mirror
While what happened in your class was pretty clearly cheating (sale of results by one student), I'm bothered by similar stories I hear in which several students collaborate on methods, and tend to converge on similar (though usually not identical) code.
IMO, this second case is a valuable lesson in the power of group development, sharing ideas, and the open source methods described by Eric Raymond in The Cathedral and the Bazaar. Much more valuable than the typical function-and-methods lessons taught in CS programs. Education has the problem of trying to assign credit for work performed, but an enlightened instructor could probably work out some compromise method.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
Jamie specifically noted that only one of several developers responded to his queries:
Getting any developers to talk about this bug has been like pulling teeth. Only one of the developers I contacted (repeatedly) even bothered to return my email.
(Maybe they've never heard of Slashdot?)
This is consistant with several other reports -- I'd also exchanged email with Blake Ross. Two emails (one asking if he'd posted to Kuro5hin) netted a grand total of "that's me :)" in response.
While I haven't seen compelling evidence of skulduggery in this case, Netscape/AOL have been less than forthcoming.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
Of the options you've given, old-style BSD. I'd like to encourage forking, including proprietary forking, of MSFT OS and apps. I'd also like it to be impossible to convert or merge it to GPL. No copylefting here.
Yes, this is punitive in intent.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
...and what's to keep you from upping the ante on the computational difficulty. Say your problem is "brute force this hash". Increasing the keylength by one byte doubles the (mean) compute time. Make this user configurable, or if you're really smart, code the program to generate problems which require some specified mean time to compute over recent requests recieved, or such.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
Unfortunately, the USPS's change of address form feeds directly into a database which is, you guessed it, sold to marketing firms.
The turnaround time for propogating the information out is about three to six months, so for the first half of a year, you're ok, but afterwards, you've become identified as a high-quality (recent) address, and to boot, recently moved so a good candidate for household goods, services,....
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
You've still missed the point. Minix isn't released with the advertising clause. That, my dear sir, is the point.
Feh!
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
I noted that the Minix license, as quoted, does not include the advertising clause. What part of this do you not understand?
I referred to the change notice posted by UC Berkeley because I couldn't find a copy of the full text of the earlier version of the BSD license. However, it is the same clause 3 which effects the advertising clause. You'll note (if you bother to follow the two thoughtfully provided links) that this clause three isn't present in the Minix license.
Minix was never issued under the prior version of the BSD license, so no, the change notice doesn't affect Minix. You appear to be confused on this point.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
While having a useful teaching tool is a laudable goal, there's no reason why the project couldn't fork, with one version of the kernel being a simple, understandable kernel, and the other serving as the basis for more complex (and presumably useful) designs.
Both the BSD and GPL licenses make forking possible. What the GPL does is ensure that, to the extent anyone wishes to do so, there are no legal embarrments to merging forks down the road. If a full-featured Minix (Maxix?) were to implement some trick-cool concept, that feature and it alone, could be merged back into Minix.
While fork/remerge is possible with BSD code which remains BSD licensed, proprietized development based on BSD code cannot be remerged back to the original core without explicit permission of the new copyright holder.
IMO Tanenbaum's probably too little, too late, though it would be interesting to see what shows up under system software at SourceForge over the next few months....
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
The license is quoted in full at the Deja archive. The advertising clause afflicted only the "old" BSD license, the language is included in the notice which rescinded this clause.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
Ulterior motives are fine, so long as there's a balance of them among those who hold power. It's when everyone's got the same ulterior motive that I start worrying.
My ulterior motives are derived from free software. You gotta problem w'dat?
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
First it's just one screen gone bad. Now it's spreading to Slashdot. Someone tell Sam.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
Dude, what are you doing posting meaningful stuff on Slashdot for?
Yes, a university which respected its student body would refer a situation such as this to its academic judiciary. It seems from the facts presented that this isn't the case here -- the school doesn't respect its students enough to supply what's rapidly becoming base-level essentials. They're going ballistic for creative appropriation of what is already freely available. Excellent suggestion, snowball's chance in hell.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
NT qualifies under UNIX98 branding as a Unix system when running the compatibility overlay previously known as Interix/OpenNT. Covered in this ZDNet article. MVS has also qualified for Unix branding, IIRC.
Not that I think of NT as Unix.
What part of "Gestalt" don't you understand?
Scope out Kuro5hin
Cookies are broken. They've outlived their usefulness, and are hopelessly open for abuse.
I have two suggestions:
The first suggestion would allow cookies to be used to track navigation and state through a single session at a site. The functionality is already available in a browser such as Netscape Navigator if you link your cookie file to /dev/null (Linux/Unix) or to a directory (Windows). Cookies are accepted but not permanently stored on your system. The upside is that cookie-dependant features of sites work. The downside is that state such as user ID and passwords have to be re-entered for each browser session.
PKE/CRA would work based on public/private key pairs, as with PGP. A user could generate as many or few of these key pairs, and optionally share them (both public and private) with other users, as desired. On entering a site requiring registration, the user could choose the key (the session identity) to send the site. If a private, secret identity is chosen, the session is personal. If a generally known key (say, cypherpunks) is sent, the session is authenticated, but not private. The remainder of the session is transacted over secure links (SSL), and cookie or other state-tracking could be used to register and/or log activity.
The strength of this scheme is allowing a user to specify both the degree of authentication, and identity authenticated used when browsing sites. If desired, keys could be generated and destroyed on a regular basis, reducing the utility of any tracking of keys. Control over whether to authenticate, who to authenticate to, and who to authenticate as, is left to the user.
Existing browser technology has been driven very strongly by server-side interestes -- user tracking, profiling, and e-commerce vendor desires. The interests of the user have not been represented, and are only partially filled by such patches as IDcide and Junkbuster (I'm another satisfied JB user). We've got the source, and with it the ability to reclaim the power.
What part of "Gestalt" don't you understand?
Interesting observation. I guess in a way SAS is something of a Perl equivalent for the mainframe -- general, high-level, all-purpose, interpreted tool. Comparatively it's relatively inexpensive (though by no means cheap), as well.
There are other MVS scripting tools though -- JCL has certain capabilities. REXX is also there, though I believe it's a relative newcomer. SAS's advantage is that it eats many common MVS file formats with relative ease. The same file-based inter-process communication method is used in both SAS programs and MVS jobs. Given that many MVS processes are heavily task, batch, and data oriented, it's a reasonably good fit.
The main differences on Unix are that there is much more system interactivity -- jobs can be interactive or trigger interactive events -- and interprocess communication often occurs as a data pipe rather than a disk file. A data step or SQL view approximates the latter. Pipelines save scads of disk I/O, and are a really cool feature. If an entire set of processes can be pipelined (data streamed from beginning to end), processing can be much faster than if a file or dataset has to be generated and read at each stage.
Unix is script-tool rich -- shells, sed, awk, Perl, Python, Tk,.... Yes. While some scripting tasks could be handled by SAS, you'd in general try to avoid doing this.
What part of "Gestalt" don't you understand?
SAS's announcement is really great news for Linux, and it's something that users had been requesting for years, going back to a request by Christa Keil in 1993. There is no single solution providing the depth and breadth of data analysis and processing capabilities as SAS. I spent seven years as a SAS developer for clients in healthcare, banking, policy research, finance, biotech, and pharmaceutical sectors.
Two years ago I launched the SAS4Linux website and mailing list to focus the campaign. While it did generate interest from users, and responses from SAS Institute, I tend to think SI arrived at its own conclusions in deciding to market the port.
As others have noted, there is a bit of "too little, too late" from a number of folks as well. I now count myself as a former (recovering?) SAS programmer. I've been sufficiently infected with free software concepts to feel that there are several things about SAS which could be improved. In fact, many of them have been, though they are approached in a number of different, and distinct, open source projects.
The obvious next step is to provide some duct tape and wrapping around these tools to provide similar functionality (though not run-alike compatibility) to SAS. Stay tuned ;-)
What part of "Gestalt" don't you understand?
As I read the announcement, the port was for "the SAS System". Presumably this includes all system components.
What I understand of the SAS architecture is that it is largely platform independent, with the dependent stuff largely isolated into a (small?) number of modules. Probably less than 10% of the codebase. Meaning that you get a lot of portability bang for your programming buck.
The SAS C Compiler is what SI themselves developed to build SAS itself. So it was probably one of the earlier technical ports to the Linux platform. Moreover, it's been generally known for a couple of years that SI have been running a compile farm based on FreeBSD, particularly for their Mainframe products. Smoke that.
My guess is that SAS/Toolkit (the developers library) will be availble for Linux -- at a price. C'est la vie.
What part of "Gestalt" don't you understand?
Which then raises the obvious question: are the 1201 anti-circumvention protections copyright-like protection of expression, or patent-like protection of ideas, and if the latter, how are they affected by decisions in MAI v. Peak and Sony v. Connectix?
What part of "Gestalt" don't you understand?
The irony is biting, particularly as I just went through a round of emails with Rob last week regarding off-topic meta-commentary on Slashdot in various posts. I'd run across an announcement of the Slashdot/VA merger in an unrelated article -- prior to Slashdot's posting notice of the announcement -- which had been moderated "Offtopic". I contested the rating, and the general practice of down-moderating Slashdot-topic posts, see below for salient commentary.
Date: Wed, 16 Feb 2000 08:32:55 -0500 (EST)
From: Rob Malda <malda@slashdot.org>
To: kmself@ix.netcom.com
cc: cmdrtaco@slashdot.org
Subject: Re: Metamoderation comments
On Tue, 15 Feb 2000 kmself@ix.netcom.com wrote:
> While the moderation is technically true, the fact is that there is *no*
> place on Slashdot to address news or issues concerning Slashdot itself.
> While the post is off _topic_, I feel it is on _meta_topic, and that
> the moderation was inappropriate.
I think its completely offtopic and deserves to be moderated down.
People need to chill out and wait until a story comes along *about* Slashdot
and talk *there* about it. At least then (surprise) I might read it and do
something about it.
--
| Rob "CmdrTaco" Malda | Just want to be misunderstood,
| malda(@)slashdot.org | I wanna be feared in my neighborhood.
| http://slashdot.org/ | Just want to be a moody man,
| cartoonist*lovemachi | say things that nobody can understand.
| ne*obsessivecompulsi | --Pete Townshend, Misunderstood
My point: Slashdot metacommentary is always, if not on topic, at least on metatopic. What's particularly ironic in this case is that Rob could have created an article to post his rant to, and ensuing commentary, but instead chose to post it here, despite what he'd said to me last week.
It's not the off-topicality that I'm objecting to in Rob's rant, it's the hypocrisy.
And I continue to request that there be a persistant, front-page linked, Slashdot-on-Slashdot forum where these conversations are on topic. Failing to provide such a forum, while encouraging the marking of such posts off-topic, is a subtle form of censorship.
What part of "Gestalt" don't you understand?
Mmmmph. I don't know about this. Is it substantially different from using a +1 score threshold if you don't want to see the "unvetted" AC comments? I suppose one might miss less with your solution, but instructing moderators to focus on raising good AC posts above 0 and increasing the number of points + moderators available might alleviate that.
Possibly, depending on the implementation.
If AC posts are relegated to "purgatory" rather than the main message board, and access to purgatory were limited to registered users, registered users of sufficient karma, or active moderators, then the positive reinforcement cycle of trolling (look at what I did to the board) is lost. ACs can't see their own trolls, and the larger public doesn't see them either. Not that the AC boards can't be spammed, but the psychological reward of doing so is greatly reduced.
Technical protections (post lockouts, redundancy filters, etc.) may still be required, but the main board problems should be much reduced.
Also agreed. Along those same lines, and somewhat aside, I'd like to see stories be assigned more than one topic (eg a story about SGI supporting Linux doesn't just get in the "Linux" category or just the "SGI" category.) Along with which users should be granted more flexibility in filtering what stories they see on the main page.
That's getting toward keywording rather than departmentalizing stories. IMO it's a good idea, and becomes necessary when either traffic or classification schemes become sufficiently rich. Face it, the world ain't heierarchical.
What part of "Gestalt" don't you understand?
There are two reasons I disagree with the "off topic" classification:
I have a serious problem with Slashdot community threads being marked off-topic, as they generally emerge in response to a blatantly evident problem, and no alternatives exist. Moderating these threads down shows an overly literal reading of the moderation guidelines, IMO.
What part of "Gestalt" don't you understand?
There would be a link between a user and an anonymous post in the database, which would allow that information to be subpeonaed into court, should it be required. That defeats entirely the whole philosophical point behind anonymous postings.
Sorry, I don't follow.
There is no link between the identity of the AC posting, and the comments posted by the AC. There is a link between the Slashdot ID of the person signing the post, and the post, but this implies absolutely nothing WRT origin of content.
What specific legal threats do you see as originating from the signing of anonymous posts to allow general viewing? AFAIK, there is no transitivity of libel from the person writing it to a person who allows it to be generally posted. The original poster's anonymity is preserved.
What part of "Gestalt" don't you understand?
There is a "hidden" forum at sid=moderation on the topic of moderation. While useful, comments posted to it have the unfortunate tendency to disappear after time.
There are obviously several very broken things about /. which are defeating the moderation system. These include troll posts, the apparent inability to restrict nonsensical or repeated posts, and an serious shortage of moderation points to posts. With karma 44, I've had moderation privileges once this millenium -- prior to December, 1999, it seemed I had mod priv typically once or twice a week. Comments get caught in a vicious cycle of failing to be moderated up once an article topic has reached a hundred posts or more -- for meaningful use, I'm forced to read at score=1, score=2, or higher.
Repeating comments I've made (and have since slipped from) sid=moderation:
Essentially, the idea is that AC posts would require vetting by at least one registered user. The user would face the karmic consequences of any moderation of the post, and would grant the AC post any karmic benefits: If the post is moderated up or down, the signer's karma is increased or reduced, and whatever default posting level the signer has is given to the post. This preserves anonymity while allowing a modicum of control over the AC process.
To do this, there would have to be a seperate viewing field for unvetted AC posts. Once vetted, the posts would enter the main forum and be subject to moderation. Only one vetting would be required to transfer the post to the main board.
There are other problems. Karma whoring, gullibility traps (posts written to look informative but actually false), etc. I believe that as other issues in the moderating system are dealt with, the magnitude of these issues will be diminished. Or they won't <g>. But there are bigger fish to fry first.
What part of "Gestalt" don't you understand?
The idea of an RBL type system is something I've thought of independently. It seems attractive. Like the UDP and real RBL, it could be a loose affiliation, decentralized, and advisory in nature. No need to bring the government in -- little that it could likely do anyway.
Realistically, what would be required is for a given network gateway to monitor its peer and child connections. Portscanning might not be necessary, depending on the signatures of an attack. A particular peer/child which exhibited behavior indicative of compromised host(s) could be blocked off, with appropriate messages sent to administrative contacts.
At the ISP level, this would include monitoring both individual dialup/fixed IP hosts, and connections to other IP aggregators. A sufficient level of filtering/blocking would act like a circuit breaker -- portions of the net might be slowed or cut off, but global abuses of the sort experienced in the past few weeks would be avoided.
What part of "Gestalt" don't you understand?
One thing I haven't seen in any news stories or most of the commentaries posted is what specific hosts and operating systems are being compromised. There was the withdrawn story to Computer Currents yesterday which claimed only Linux and Solaris were involved. I find this hard to believe. I've heard anectdotal evidence that Windows machines are the most frequently compromised hosts, via viruses.
If the truth is lurking somewhere in earshot, could it please make itself heard?
What part of "Gestalt" don't you understand?