SCO boots boss McBride ( Linux lawsuits live )

The Register ^ | 19th October 2009 18:50 GMT | Austin Modine

[altair]

But that the thing was created in c and uses the cout/cin[/cerr] model from the stdio header, would it have been overly difficult to “snip-out” the globbing into, say, a globbing header? If that HAD been the case then I would wager that ALL [non-toy] shells would have avoided the problem.

It never entered into the SVID and later the POSIX specs. I'm not sure why.

As for shell-scripts... I have mixed feelings on them, generally I’d say they’re bad because of the dependencies they pick up from version to version & flavor to flavor of unix/linux; I’m of the strong opinion that high-level languages should allow you to, as much as possible, work with general disregard to the underlying hardware-architecture.

It's not particularly difficult to write portable shell scripts nowadays. Things were a lot different in the wild '80s when there were dozens of different flavors of Unix around including things like Eunice. In my work environment I can assume that /bin/zsh exists, so I tend to write anything complicated with zsh extensions.

What about the wonderful [/sarc] allowance of assignments in the condition-test for control structures?

It depends upon the language. Check out the language formerly named Icon some time for a valid use case.

Ada’s endif which prevents the dangling-else problem.

We were allowed to latitude (another learning experience in and of itself) to define our own grammar. I was also involved with Modula II at the time and hated that construction as a programmer until I saw how it simplified the compiler and added it to my own project.

Another Ada feature that I like is the Array’Range looping; if you use that you’ll never access an invalid index in that array

More important was making the index variable local to the scope of the for. There are use cases that this prevents and none of them are all that good.

Are you using GNAT, by the way? That implementation has gotten very good and is better now than any of the commercial Ada compilers I used in the late '80s.

Well, I was thinking if it was handled at the OS-level, well defined, and outside-linkable like c’s stdio (I’m sure I could think up a better example) {maybe outside-interfacable would be stating it better} would make it a lot easier to work-out... that is grow it with the OS.

Been there, done that, still carry the scars. It's a beautiful, beautiful rose that smells bad. The commercialized work never went anywhere.

>Do I take it that you’re a fan of ASN.1 or XDR?

I haven’t actually heard of them. Why do you think I’d be a fan?

Those are the only two remotely successful vendor neutral binary methods of passing data between applications. ASN.1 was less successful than XDR (Sun RPC). I was just checking to see how much you were influenced by Microsoft vendor lock-in solutions.

I’m somewhat disappointed in the state of OSes right now.

I'm not, but then I've worked most of my adult life to achieve what has become a "modern" Linux distro. I'm hoping to pass down to them my youthful enthusiasm and see where it takes them.

>Most compilers today use LR techniques not LL techniques and filling in error states in LR tables is something I’ve never seen well-automated.

LL v LR... I’ve heard of them, but haven’t really investigated. I probably should, but I’m getting ready to graduate. (So mayhaps after.)

Hmmm. Your compiler class should have discussed that. Oh well. It's mostly academic now.

Do you have a job lined up for after graduation? I can offer you an introduction to my company, though it would likely require you to relocate to North Carolina. FReepmail me if you're interested.

[OneWingedShark]

>>What about the wonderful [/sarc] allowance of assignments in the condition-test for control structures?
>
>It depends upon the language. Check out the language formerly named Icon some time for a valid use case.

I’ve heard of Icon; it was designed by the same guy that designed SNOBAL, IIRC. (SNOBOL looks rather interesting, I just read the 5 or 6 page pdf of the [design] paper on it.)

I agree that it depends on the language, more specifically the philosophy if it; but it’s also interesting to note that the assignment in Icon seems to act like a control structure itself. (Glancing at the brief intro.)

>>Ada’s endif which prevents the dangling-else problem.
>
>We were allowed to latitude (another learning experience in and of itself) to define our own grammar.

Oh, quite agreed. Sadly we weren’t; we had to do a C knock-off.

>I was also involved with Modula II at the time and hated that construction as a programmer until I saw how it simplified the compiler and added it to my own project.

Nifty. I went the other way, looking at the grammar/usage of the endif I saw how it added better definition to the program and thereby reduced typo-style errors.

>>Another Ada feature that I like is the Array’Range looping; if you use that you’ll never access an invalid index in that array
>
>More important was making the index variable local to the scope of the for. There are use cases that this prevents and none of them are all that good.

Oh, I quite agree with you there; but all the “use the index after the loop to modify that last element” uses are laid to rest with Array’First and Array’Last.

>Are you using GNAT, by the way? That implementation has gotten very good and is better now than any of the commercial Ada compilers I used in the late ‘80s.

Yes [and no]. I downloaded it, and have played around with it a little, but haven’t yet done anything of note. I’m still reading the ADA 2005 book by John Barnes; {ISBN: 0321340787} to learn/familiarize myself with the language a bit before really starting on things. (Though I’m from a Delphi/Pascal [self-taught] background so it’s not a tough transition.)

I like a lot of Ada’s compile-time checking, and the tasking-as-part-of-the-language seems far superior to the Cuda/OpenML/other-parallel-extensions-or-APIs that I’ve tangentially seen in C/C++.

I see how such facilities could be VERY useful in OS-Design. {Of course there’s always the bootstrapping problem, but that’s logically unavoidable.}

What are your experiences with Ada? Generally good, bad, or indifferent?

>>>Do I take it that you’re a fan of ASN.1 or XDR?
>>
>>I haven’t actually heard of them. Why do you think I’d be a fan?
>
>Those are the only two remotely successful vendor neutral binary methods of passing data between applications.

Ah, I see. Though isn’t CORBA also a vendor-neutral method for doing so too... granted it’s more targeted to machines over networks.

>I was just checking to see how much you were influenced by Microsoft vendor lock-in solutions.

LOL - I’m not sure I am locked in. Though I’ll say that I do tend to like my windows box more than the university’s Linux boxes... I always feel like the Unix/Linux OSes are working against me (most especially in CLI mode). But then again, I also like Win98SE more than I do Vista.

.NET seems to be an interesting idea, but there seems to be a design-flaw in that you have to have a directory for each version... a versioning-tag would have been better IMO. I’m not particularly a fan of C#, which MS seems to be aggressively pushing (especially with its XNA game-dev).

Most of the people in mt CS dept are [rather hardcore] anti-microsoft; while I don’t agree with everything MS does, I have to say that they HAVE done a lot for standardization in CS. {True there are some standards MS doesn’t comply with, but that’s a typical NIH (not invented here) failing.}

But one thing that bugs me about my CS dept is the standard answer/reply of “just download the linux source-code” to finding out I would like to make an OS.

>>LL v LR... I’ve heard of them, but haven’t really investigated. I probably should, but I’m getting ready to graduate. (So mayhaps after.)
>
>Hmmm. Your compiler class should have discussed that. Oh well. It’s mostly academic now.

Should have but didn’t; probably the same reason we weren’t given leeway on the language-design: it was just a 1-semester class.

>Do you have a job lined up for after graduation?

Nope; not really. I have a friend popping a resume to his superiors and that’s about it right now.

>I can offer you an introduction to my company [snip] FReepmail me if you’re interested.

Thank you much. I think I will. (Though I’ll be honest when I say I’m not terribly fond of the idea of moving; *shrug* you got to do what you got to do, right?)