Yes, I agree! Photoshop is the worst offender.
on
Version Fatigue
·
· Score: 2
As soon as I read this story I immediately thought "Photoshop". With every version (the exception perhaps being 7, which I am really liking) they change around the key commands for things that have always existed, especially the pencil and paintcan. They also love to move stuff around in their menus, and change the key shortcuts for those, too. I don't think it even has to do with complex new features -- just some kind of randomized user study where they decide that it really makes more sense for the pencil to be 'b' instead of 'y', or for crop to be under 'image' instead of 'edit'.
I'm ok with changing the basic interaction with the software (type tool, choosing brushes, etc.) if they want to introduce new features, but changing the key commands for tools and operations that have already been there is totally insane!!
Use robots.txt, stupido. It lets you prevent search engines from indexing and archiving your property. However, if you're that concerned about people copying your pages, you might try avoiding the internet.
I personally love the internet archive and google's cache.
Your needs for parsing regular (or regular-ish) grammars in perl will be filled. Unfortunately, regular expressions are not always the right tool for the job (see the gargantuan e-mail address parser in the back of the MRE book for an example) because many grammars are not regular. Try parsing (say) C with perl! Also, perl has poor facilities for manipulating such parse trees, especially compared to a language with datatypes and pattern matching like SML or O'caml.
Perl makes parsing flat data brief (though regular expression libraries in other high level languages make it just as easy, if a little more verbose), but it is not by any stretch the be-all and end-all of parsing.
While it's kind of scary that you're a senior with 7 years of experience and don't know this, it should be easy to catch up by reading the beginning of a compilers book. Older compiler books tend to concentrate on parsing a lot, since parsing used to be "most" of the work involved in compiling. Actually, all the theory on how to conserve the last bit of memory using some kind of wacky table is pretty useless these days, but it's there.
Though perl's built-in regexp support makes its syntax more brief for matching regular grammars, many grammars are not regular (The grammars of most programming languages are context free or worse!). For those grammars, a parser generator like bison can make life a bit easier.
Then, manipulating parse trees in a language like C or perl is pretty painful once you've used datatypes and pattern matching in a language like SML, O'Caml, or Haskell. If you're going to be doing a lot of work on the parsed data, and its structure is not trivial, give these a shot.
Finally, about the only thing that's excited me about parsing for years is the idea of parsing combinators: writing parsers that look like the grammar IN the language itself. They can allow the parsing of arbitrary grammars, and don't require resorting to external tools. There's a good beginning explanation in Paulson's "ML for the Working Programmer", though I can't find a tutorial online... maybe it's time to write one!
It's true that apache has a pretty good track record, but sadly this bug was not found through code review or anything like that. It was found using the same means available against a closed-source server: spamming the thing with malformed requests. I'm not sure a security issue is "obscure" if it's in the default configuration, either...
> For every daemon running with 'nobody' permissions (or any shared 'daemon' UID), the > security risk increases exponentially, as a flaw in one daemon means access to the process > data of all of the others.
If n is the 10,000 elements (that seems like the obvious thing), then this certainly won't be lg(n). I suppose you could be counting something like the number of character comparisons, but if the strings are 10k each, then a sorted list REALLY isn't the right dictionary data structure.
Well, I didn't bother reading the article (because hey, this is slashdot), but I must say that I've actually had really good success convincing my intermediately-computer-skilled friends to use Mozilla. They don't have any sort of browser loyalty. I just showed them tabbed browsing with middle click, ad blocking, the sidebar, and they decided that it was better than IE.
That, and, I hear AOL is switching to NS6?
Anyway, why does it matter? Mozilla has enough users and developers now to support it as an open source project, so winning or losing the browser "war" doesn't mean much to those of us who just want a good browser to use..
Netscape 4 on linux had an exploitable hole in their JPEG decoder. That is, a specially crafted JPEG could be used to execute arbitrary code on the target's machine. Could that code then "infect" other JPEGs? Sure. Would it actually spread? No, but if there were a similar bug in the default windows JPEG viewer, it wouldn't be surprising at all to see a similar worm spread.
http://www.openwall.com/advisories/OW-002-netsca pe -jpeg.txt
(I recall that this bug was successfully exploited; that advisory seems more tentative..)
> "Their ideas are not open to all, and they strive for competetive ends which do not benefit all of mankind, only M$ and it's loyal ass-kissers."
This is simply not true. They publish their ideas in scientific journals, just like other research labs, and publish open source software. It's true that some of their research is oriented towards Microsoft products, but MS itself hardly ever actually adopts MSR ideas. If you think I'm wrong, point me to some MSR technologies that have been patented and then used exclusively by MS, or something like that.
I'm reacting to the posts in this article, many of which are knee-jerks like "ANOTHER language from Microsoft! Who needs more languages??" It's true that lots of languages run on linux (though most of the ones you list are C or safe-C variants), though I don't think there is much enthusiasm for *new* programming language ideas on linux. That's pretty sad, because the linux folks are the ones who are supposed to adopt ideas because of their technical value!
He's right. There's no way that people are going to get excited about this -- for most people, renting a DVD is much simpler than getting ahold of a DivX CD. (Especially if they don't have a computer!!)
Also, don't be surprised if this community of modders gets a big fat DMCA lawsuit. This is EXACTLY the kind of activity that the DMCA was created to make illegal. (It's a crappy law, for sure, but there it is...)
Once again, you guys are confusing Microsoft Research with Microsoft. It's a different place. They don't put out crappy, buggy software, or try to take over the world. It's just a research lab.
Incidentally, MSR Cambridge has some of the best programming language researchers in the world! (Personally, I wish the linux world had more tolerance for new PL ideas..)
F# is not crippled. It's just missing the module constructs of O'Caml, which as far as I know should not cause any problem to be implemented -- they just aren't yet.
SML.NET, which I think works (though I haven't checked) apparently implements all of SML, including its module language. SML's module language is even cooler than O'caml's, too.
Slashdot Mirror
I agree with you, and in general you can sum up your argument as follows: Software sucks because we use inappropriate abstraction mechanisms.
It had better not be filled with hydrogen.
As soon as I read this story I immediately thought "Photoshop". With every version (the exception perhaps being 7, which I am really liking) they change around the key commands for things that have always existed, especially the pencil and paintcan. They also love to move stuff around in their menus, and change the key shortcuts for those, too. I don't think it even has to do with complex new features -- just some kind of randomized user study where they decide that it really makes more sense for the pencil to be 'b' instead of 'y', or for crop to be under 'image' instead of 'edit'.
I'm ok with changing the basic interaction with the software (type tool, choosing brushes, etc.) if they want to introduce new features, but changing the key commands for tools and operations that have already been there is totally insane!!
Use robots.txt, stupido. It lets you prevent search engines from indexing and archiving your property. However, if you're that concerned about people copying your pages, you might try avoiding the internet.
I personally love the internet archive and google's cache.
Your needs for parsing regular (or regular-ish) grammars in perl will be filled. Unfortunately, regular expressions are not always the right tool for the job (see the gargantuan e-mail address parser in the back of the MRE book for an example) because many grammars are not regular. Try parsing (say) C with perl! Also, perl has poor facilities for manipulating such parse trees, especially compared to a language with datatypes and pattern matching like SML or O'caml.
Perl makes parsing flat data brief (though regular expression libraries in other high level languages make it just as easy, if a little more verbose), but it is not by any stretch the be-all and end-all of parsing.
While it's kind of scary that you're a senior with 7 years of experience and don't know this, it should be easy to catch up by reading the beginning of a compilers book. Older compiler books tend to concentrate on parsing a lot, since parsing used to be "most" of the work involved in compiling. Actually, all the theory on how to conserve the last bit of memory using some kind of wacky table is pretty useless these days, but it's there.
Though perl's built-in regexp support makes its syntax more brief for matching regular grammars, many grammars are not regular (The grammars of most programming languages are context free or worse!). For those grammars, a parser generator like bison can make life a bit easier.
Then, manipulating parse trees in a language like C or perl is pretty painful once you've used datatypes and pattern matching in a language like SML, O'Caml, or Haskell. If you're going to be doing a lot of work on the parsed data, and its structure is not trivial, give these a shot.
Finally, about the only thing that's excited me about parsing for years is the idea of parsing combinators: writing parsers that look like the grammar IN the language itself. They can allow the parsing of arbitrary grammars, and don't require resorting to external tools. There's a good beginning explanation in Paulson's "ML for the Working Programmer", though I can't find a tutorial online... maybe it's time to write one!
The Black Mesa Federal Research Facility had a maglev train, at least, until it was destroyed by aliens from planet Xen.
It's true that apache has a pretty good track record, but sadly this bug was not found through code review or anything like that. It was found using the same means available against a closed-source server: spamming the thing with malformed requests. I'm not sure a security issue is "obscure" if it's in the default configuration, either...
Were you sleeping last night? Last week? Apache wasn't secure then, either. And it probably won't be even after you apply the patch.
> For every daemon running with 'nobody' permissions (or any shared 'daemon' UID), the
> security risk increases exponentially, as a flaw in one daemon means access to the process
> data of all of the others.
That's not exponential, that's quadratic.
If n is the 10,000 elements (that seems like the obvious thing), then this certainly won't be lg(n). I suppose you could be counting something like the number of character comparisons, but if the strings are 10k each, then a sorted list REALLY isn't the right dictionary data structure.
You need to put special Microwave "power beams" in the ceiling, and beam the power to the laptops. Put these all over the hospital.
Well, I didn't bother reading the article (because hey, this is slashdot), but I must say that I've actually had really good success convincing my intermediately-computer-skilled friends to use Mozilla. They don't have any sort of browser loyalty. I just showed them tabbed browsing with middle click, ad blocking, the sidebar, and they decided that it was better than IE.
That, and, I hear AOL is switching to NS6?
Anyway, why does it matter? Mozilla has enough users and developers now to support it as an open source project, so winning or losing the browser "war" doesn't mean much to those of us who just want a good browser to use..
> Fine an element in this sorted list (of say 10000)
> Would write a n order search instead of a binary search.
And how do you propose to do sub-linear lookup in a LIST?
Fibonacci is really, really easy:
int n1 = 0, n2 = 1, t;
for(;;) {
t = n1 + n2;
printf("%d ", t);
n1 = n2;
n2 = t;
}
Ackermann requries an explicit stack (it's not primitive recursive), but it's also pretty easy.
Someone's cousin's brother is also their cousin.
Netscape 4 on linux had an exploitable hole in their JPEG decoder. That is, a specially crafted JPEG could be used to execute arbitrary code on the target's machine. Could that code then "infect" other JPEGs? Sure. Would it actually spread? No, but if there were a similar bug in the default windows JPEG viewer, it wouldn't be surprising at all to see a similar worm spread.
a pe -jpeg.txt
http://www.openwall.com/advisories/OW-002-netsc
(I recall that this bug was successfully exploited; that advisory seems more tentative..)
> "Their ideas are not open to all, and they strive for competetive ends which do not benefit all of mankind, only M$ and it's loyal ass-kissers."
This is simply not true. They publish their ideas in scientific journals, just like other research labs, and publish open source software. It's true that some of their research is oriented towards Microsoft products, but MS itself hardly ever actually adopts MSR ideas. If you think I'm wrong, point me to some MSR technologies that have been patented and then used exclusively by MS, or something like that.
The filename is irrelevant; md5sum just uses the contents of the file.
Or are you doing:
echo secret | md5sum
??
Wow... 68,000 lines of C++ for the engine and a mere 5,300 lines of "high-level" code is NOT what I'd call software engineering. ;)
I'm reacting to the posts in this article, many of which are knee-jerks like "ANOTHER language from Microsoft! Who needs more languages??" It's true that lots of languages run on linux (though most of the ones you list are C or safe-C variants), though I don't think there is much enthusiasm for *new* programming language ideas on linux. That's pretty sad, because the linux folks are the ones who are supposed to adopt ideas because of their technical value!
He's right. There's no way that people are going to get excited about this -- for most people, renting a DVD is much simpler than getting ahold of a DivX CD. (Especially if they don't have a computer!!)
Also, don't be surprised if this community of modders gets a big fat DMCA lawsuit. This is EXACTLY the kind of activity that the DMCA was created to make illegal. (It's a crappy law, for sure, but there it is...)
Once again, you guys are confusing Microsoft Research with Microsoft. It's a different place. They don't put out crappy, buggy software, or try to take over the world. It's just a research lab.
Incidentally, MSR Cambridge has some of the best programming language researchers in the world! (Personally, I wish the linux world had more tolerance for new PL ideas..)
I think he's talking about a computer or something, I can't figure it out.
F# is not crippled. It's just missing the module constructs of O'Caml, which as far as I know should not cause any problem to be implemented -- they just aren't yet.
SML.NET, which I think works (though I haven't checked) apparently implements all of SML, including its module language. SML's module language is even cooler than O'caml's, too.