For me, Freeciv is not only more playable than Civ II, it is also prettier (for me, the Civ II graphics and distractions get in the way of gameplay). It does look its age, but so does chess.
Perhaps you should try playing some Freeciv and report your findings.
These comments (like all comments I've seen on Microsoft) seem to underestimate the effort it takes to maintain backward compatibility. This is a huge problem. They have finally introduced a serious, state of the art programming environment (.NET) but how to use it to their advantage? If only to avoid bad publicity they'll have to maintain their existing stuff until eternity. What's the challenge in writing a decent word processor for.NET if you know you can only market it when it's 100% Word-compatible? Open Office has that problem, but Microsoft has it just the same.
I have really thought on how to get strong typing in Perl 5, and i think it's impossible. With strong typing I mean full compile-time verification that wherever a variable (or constant, parameter, array slice, any piece of syntax that denotes a value) is asserted to be of a certain type or in a certain class, it is 100% certain that all values this variable can possibly assume will indeed be of that type / in that class.
For arbitrary Perl 5 code this is obvious, but I don't want it for arbitrary code: I want it for my own Perl code. I want to be able to provide type/class definitions for all the arguments and return values of all subroutines that appear in a particular piece of Perl source code (both my own and calls to subs defined elsewhere) and have the compiler verify: assuming that all these foreign subroutines satisfy the typing specified here, are my own subroutines 100% guaranteed to satisfy it as well?
There doesn't seem to be a way to do this with Perl. Perhaps there is something in the Devel::* namespace now which is suitable? It's been a while since I tried this (and gave up on programming larger projects in Perl for that reason).
But they do have a decent security procedure, an official from the dept. of Justice explained on TV. It just wasn't followed in this case. The guy shouldn't even have had this stuff on a PC at home. Never attribute to lack of brueaucracy what can be accounted for by arrogance alone.
i don't buy your argument. indeed, many documents are called FAQs because they are about a product or service and are in the Q?A form, and yes, many people (especially their authors I suspect) consider the meaning of the term FAQ to coincide with this use. but I have never agreed. have such documents, by all means, they can be vey useful, but call them by different names! a FAQ should literally be a list of *actual* questions and actual answers to them - there is often a great need for just such a list.
Frankly I have grown disappointed with the level of reporting on software from Slashdot. A main reason for me to prefer open source software over commercial software was that what you got to read about it came from real developers and real users of real, readily available software. Since it focused on open source software, Slashdot's reporting used to reflect this, announcing release reports from actual developers on actual releases of software that people actually used.
Today many of the "news" items on software releases that feature on Slashdot are no longer on actual releases, but announcements on future releases, delays on future releases, plans on future releases, etcetera. The announcers are not developers but CEOs, marketeers, magazine columnists, tcetera. Consequently the "news" items themselves and the ensuing discussions are shrouded in marketese and speculation, and generally demonstrate a very superficial, PC-ish outlook on software, treating applications or even whole OSes like participants in a sports competition. "Will Microsoft's (KDE's, Mandrake's, Enlightenment's,...) New Team Top The League Again In 2005?" Having to wade through this hogwash is what turned me off commercial software; now that sites like Slashdot and their users give free software the same treatment, both the sites and the software itself lose a major competitive advantage. Slashdot is a major culprit.
Interestingly enough, Microsoft has made a very successful move in the opposite direction by letting its developers blog on their daily work, which provides us users/programmers with the kind of communication channel that sites like Slashdot used to provide for open source software.
It would help if Slashdot introduced a system to separate advertisements, in whatever form, from real reports on real product releases.
While I can - barely - imagine a time when a caps lock key was convenient in entering long pieces of written text natural language, namely, in cases where all caps was used to make titles, keywords, etc., stand out, it should be universally recognized by now that STRINGS OF ALL-CAPS WORDS ARE HARD TO READ AND SIMPLY ANNOYING, no matter in which context they appear.
For computer keyboards this time frame must have been short: languages such as Cobol, Fortran and Algol were developed at a time when *all* text was in all caps, and when keyboards, line printers and teletype terminals with the capability to distinguish between lower case and upper case letters and line printers became commonplace, in the early 70s, text processing software was already supporting the formatting of text in other ways (italicize, enbolden, etc.)
So, in short, having a caps lock key was never really useful, and it never will be. I have never used since I moved to typing on a computer keyboard, some 22 years ago. The only 100% effective way of dealing with the caps lock key is to rip it out of the keyboard, and this is what I do as a matter of routine.
Of course they must be interpreted by computers, but the reason they exist is that humans (the author, or someone else) must be able to understand what is going on.
So it's very important to find expressions for humans to understand and use. Perhaps natural languages can have some undiscovered gems for us.
On the other hand, there are far more programming languages than natural languages already, we're quite capable of inventing new ways to say things without knowing hundreds of existing ways in advance.
OK, so you map port 80 to 8000, but how do you convince Apache not to redirect its/-less directory requests to http://my.example.com:8000/path/to/directory/ ? Apache's configuration has many arbitrary limitations in that area and you have to invoke some module from hell to get it right. Do you just enable mod_rewrite and accept the risks? I'm not willing to do that. What else, hack mod_dir to leave off its port number, so my fellow sysadmin no longer knows how to upgrade Apache?
Generally speaking there is always a tradeoff: tightening security on one point is going to complicate the mechanisms of operation, which is not only more work for all concerned, but also brings new risks to security and reliability.
let's see the *original* "three click rule"
on
Web 'Rules' Changing?
·
· Score: 2, Informative
I notice this "Testing the three-click rule" paper is well advertised in many places, but its accusation that we have a rule here that isn't backed up by any empirical research is completely ludicrous. The author apparently didn't bother to search the literature, and doesn't even appear to know the original paper.
To quote:
Directions for Design
Since users accessed on average 10 pages per server, this would indicate that "must see" information must be accessible within two to three jumps of the initial home page (two/three navigations in, two/three out, performed three/two times). However, [...]
This paper is one of the first, if not the very first, actual user survey studies on the Web. It is very limited in scope, of course, and there may be good arguments to question its validity, but if you're going to do that, at least quote the rule correctly, mention its origins, and mention the fact that it was co-written by James Pitkow, who has continuing this line of research until the present day.
It's very useful to borrow words from other languages. You create specialized terms that won't be confused with existing words or concepts. Even the language you borrow them from doesn't have them!
For example, in Dutch we have the English loan word "mail" for e-mail. It only means electronic mail, so it can't be confused with any other form of mail, for which we already had a completely different word ("post"). Every translation into "real Dutch" would require effort and produce a term that would have to be articifially forced into this specialized meaning, while the loan word "mail" comes for free, at least to those of us who speak English (which is about 90% of the users of e-mail).
A "real" translation for words doesn't make sense from a practical point of view, unless the users will have problems with the borrowed word. The use of English loan words in French increases with the fluency in English of the francophones. Trying to find "real French" translations can be useful, but insisting on them only serves purism, which is a fun game for intellectuals and politicians, but not very useful otherwise.
just as a technical point, no computer is equivalent to a Turing machine, for all have finite memory. many programming languages also assume finite memory.
caching should be a separate filesystem module, like Solaris cachefs. then you can support caching for any filesystem, not just whatever lufs happens to support.
data transfer can be optimized a lot by using rsync or something equivalent.
btw a cute idea (but based on HTTP rather than SSH) is Zero Install
Second: XML is a bracketed language, and therefore, much *easier* to parse than arbitrary regular languages, let alone context-free ones. Parsing is the activity of recognizing structure in input that doesn't explicitly encode it. This can be costly because it may require nondeterminism. In an XML document, the structure is already marked up explicitly. The structure can simply be read off deterministically.
Your comment does apply, not to the task of parsing XML in general, but to the problem of wanting to quickly determine if given content, available in a document, matches a certain pattern. This can be done with sublinear algorithms. Note that XML was not intended to be used for that purpose: it's an interchange format, not a data storage format.
Second: nothing prevents you from writing an XML document scanner that uses all the known tricks and algorithms of sublinear string matching on regular (flat) XML, even when it still handles non-regular XML (trees of arbitrary depth) correctly. So your comment doesn't point out an inherent drawback of XML, although it does apply to standard off-the-shelf XML processors.
In short, you have a point, but the conclusion is nonsense.
Incidentally, the new Freeciv release (1.14.0 came out on Jan 17th) wasn't considered newsworthy either.
Freeciv would have won more polls if it had only been nominated there. This tells us something both about the value of these polls and the state of gaming on Linux.
Slashdot Mirror
For me, Freeciv is not only more playable than Civ II, it is also prettier (for me, the Civ II graphics and distractions get in the way of gameplay). It does look its age, but so does chess.
Perhaps you should try playing some Freeciv and report your findings.
These comments (like all comments I've seen on Microsoft) seem to underestimate the effort it takes to maintain backward compatibility. This is a huge problem. They have finally introduced a serious, state of the art programming environment (.NET) but how to use it to their advantage? If only to avoid bad publicity they'll have to maintain their existing stuff until eternity. What's the challenge in writing a decent word processor for .NET if you know you can only market it when it's 100% Word-compatible? Open Office has that problem, but Microsoft has it just the same.
In any case, you don't need to learn Choktaw to appreciate this distinction. In my native tongue you use an auxiliary verb for this purpose:
De man werd opgepakt. Hij had een bank beroofd.
De man werd opgepakt. Hij zou een bank hebben beroofd.
In English you use an adverb, or some other sentence adornment; for example, to translate the two sentences above:
The man was taken in custody. He had robbed a bank.
The man was taken in custody. He had allegedly robbed a bank.
Choktaw can allegedly express this difference by modifying the form of the verb itself.
I have really thought on how to get strong typing in Perl 5, and i think it's impossible. With strong typing I mean full compile-time verification that wherever a variable (or constant, parameter, array slice, any piece of syntax that denotes a value) is asserted to be of a certain type or in a certain class, it is 100% certain that all values this variable can possibly assume will indeed be of that type / in that class.
For arbitrary Perl 5 code this is obvious, but I don't want it for arbitrary code: I want it for my own Perl code. I want to be able to provide type/class definitions for all the arguments and return values of all subroutines that appear in a particular piece of Perl source code (both my own and calls to subs defined elsewhere) and have the compiler verify: assuming that all these foreign subroutines satisfy the typing specified here, are my own subroutines 100% guaranteed to satisfy it as well?
There doesn't seem to be a way to do this with Perl. Perhaps there is something in the Devel::* namespace now which is suitable? It's been a while since I tried this (and gave up on programming larger projects in Perl for that reason).
But they do have a decent security procedure, an official from the dept. of Justice explained on TV. It just wasn't followed in this case. The guy shouldn't even have had this stuff on a PC at home. Never attribute to lack of brueaucracy what can be accounted for by arrogance alone.
Thanks for explaining damiam's meaning.
i don't buy your argument. indeed, many documents are called FAQs because they are about a product or service and are in the Q?A form, and yes, many people (especially their authors I suspect) consider the meaning of the term FAQ to coincide with this use. but I have never agreed. have such documents, by all means, they can be vey useful, but call them by different names! a FAQ should literally be a list of *actual* questions and actual answers to them - there is often a great need for just such a list.
1) ln -s libfoo.so.5 libfoo.so.6
2) pray
Thus far it's always appeared to work for me
but it isn't really *supposed* to, of course.
Frankly I have grown disappointed with the level of reporting on software from Slashdot. A main reason for me to prefer open source software over commercial software was that what you got to read about it came from real developers and real users of real, readily available software. Since it focused on open source software, Slashdot's reporting used to reflect this, announcing release reports from actual developers on actual releases of software that people actually used.
Today many of the "news" items on software releases that feature on Slashdot are no longer on actual releases, but announcements on future releases, delays on future releases, plans on future releases, etcetera. The announcers are not developers but CEOs, marketeers, magazine columnists, tcetera. Consequently the "news" items themselves and the ensuing discussions are shrouded in marketese and speculation, and generally demonstrate a very superficial, PC-ish outlook on software, treating applications or even whole OSes like participants in a sports competition. "Will Microsoft's (KDE's, Mandrake's, Enlightenment's, ...) New Team Top The League Again In 2005?" Having to wade through this hogwash is what turned me off commercial software; now that sites like Slashdot and their users give free software the same treatment, both the sites and the software itself lose a major competitive advantage. Slashdot is a major culprit.
Interestingly enough, Microsoft has made a very successful move in the opposite direction by letting its developers blog on their daily work, which provides us users/programmers with the kind of communication channel that sites like Slashdot used to provide for open source software.
It would help if Slashdot introduced a system to separate advertisements, in whatever form, from real reports on real product releases.
While I can - barely - imagine a time when a caps lock key was convenient in entering long pieces of written text natural language, namely, in cases where all caps was used to make titles, keywords, etc., stand out, it should be universally recognized by now that STRINGS OF ALL-CAPS WORDS ARE HARD TO READ AND SIMPLY ANNOYING, no matter in which context they appear.
For computer keyboards this time frame must have been short: languages such as Cobol, Fortran and Algol were developed at a time when *all* text was in all caps, and when keyboards, line printers and teletype terminals with the capability to distinguish between lower case and upper case letters and line printers became commonplace, in the early 70s, text processing software was already supporting the formatting of text in other ways (italicize, enbolden, etc.)
So, in short, having a caps lock key was never really useful, and it never will be. I have never used since I moved to typing on a computer keyboard, some 22 years ago. The only 100% effective way of dealing with the caps lock key is to rip it out of the keyboard, and this is what I do as a matter of routine.
2000 is not "before Google".
w ww.altavista.com/
The oldest Altavista page in the wayback archive,
http://web.archive.org/web/19961022174810/http://
is much smaller, but it already features ads. Early Altavista was adless.
Programming languages are for humans!
Of course they must be interpreted by computers, but the reason they exist is that humans (the author, or someone else) must be able to understand what is going on.
So it's very important to find expressions for humans to understand and use. Perhaps natural languages can have some undiscovered gems for us.
On the other hand, there are far more programming languages than natural languages already, we're quite capable of inventing new ways to say things without knowing hundreds of existing ways in advance.
I'm still compiling and playing Freeciv on my P1
(with more memory). It definiitely compiles on it.
OK, so you map port 80 to 8000, but how do you convince Apache not to redirect its /-less directory requests to http://my.example.com:8000/path/to/directory/ ? Apache's configuration has many arbitrary limitations in that area and you have to invoke some module from hell to get it right. Do you just enable mod_rewrite and accept the risks? I'm not willing to do that. What else, hack mod_dir to leave off its port number, so my fellow sysadmin no longer knows how to upgrade Apache?
Generally speaking there is always a tradeoff: tightening security on one point is going to complicate the mechanisms of operation, which is not only more work for all concerned, but also brings new risks to security and reliability.
The original source for the "three-click rule" is Catledge and Pitkow's 1995 paper, Characterizing Browsing Strategies in the World Wide Web.; see an online copy.
To quote: Directions for Design Since users accessed on average 10 pages per server, this would indicate that "must see" information must be accessible within two to three jumps of the initial home page (two/three navigations in, two/three out, performed three/two times). However, [...]
This paper is one of the first, if not the very first, actual user survey studies on the Web. It is very limited in scope, of course, and there may be good arguments to question its validity, but if you're going to do that, at least quote the rule correctly, mention its origins, and mention the fact that it was co-written by James Pitkow, who has continuing this line of research until the present day.
For example, in Dutch we have the English loan word "mail" for e-mail. It only means electronic mail, so it can't be confused with any other form of mail, for which we already had a completely different word ("post"). Every translation into "real Dutch" would require effort and produce a term that would have to be articifially forced into this specialized meaning, while the loan word "mail" comes for free, at least to those of us who speak English (which is about 90% of the users of e-mail).
A "real" translation for words doesn't make sense from a practical point of view, unless the users will have problems with the borrowed word. The use of English loan words in French increases with the fluency in English of the francophones. Trying to find "real French" translations can be useful, but insisting on them only serves purism, which is a fun game for intellectuals and politicians, but not very useful otherwise.
just as a technical point, no computer is equivalent to a Turing machine, for all have finite memory. many programming languages also assume finite memory.
don't forget lazyfs, lufs, autofs
data transfer can be optimized a lot by using rsync or something equivalent.
btw a cute idea (but based on HTTP rather than SSH) is Zero Install
First: XML is about tools and libraries.
Second: XML is a bracketed language, and therefore, much *easier* to parse than arbitrary regular languages, let alone context-free ones. Parsing is the activity of recognizing structure in input that doesn't explicitly encode it. This can be costly because it may require nondeterminism. In an XML document, the structure is already marked up explicitly. The structure can simply be read off deterministically.
Your comment does apply, not to the task of parsing XML in general, but to the problem of wanting to quickly determine if given content, available in a document, matches a certain pattern. This can be done with sublinear algorithms. Note that XML was not intended to be used for that purpose: it's an interchange format, not a data storage format.
Second: nothing prevents you from writing an XML document scanner that uses all the known tricks and algorithms of sublinear string matching on regular (flat) XML, even when it still handles non-regular XML (trees of arbitrary depth) correctly. So your comment doesn't point out an inherent drawback of XML, although it does apply to standard off-the-shelf XML processors.
In short, you have a point, but the conclusion is nonsense.
We'd probably still have the free BSDs and their tools. Not quite as extensive as GNU's but good enough to work with.
Incidentally, the new Freeciv release (1.14.0 came out on Jan 17th) wasn't considered newsworthy either.
Freeciv would have won more polls if it had only been nominated there. This tells us something both about the value of these polls and the state of gaming on Linux.
what the article refers to is Apache modules. mod_php in your case.
But does it have the heuristic parsing of wuasi-HTMLthat HTML::Parser / HTML::TreeBuilder were designed to do?
Backreferences (in Perl 4 and some other regex libraries, but not in sed or awk), can express things like
{ xxx | x in \Sigma }
which is not a type 2 language:
% cat input
ab
abc
abcabc
abcabcabc
abcabcabcabc
abcabcabcabcabc
% perl -lne '/^(.+)\1\1$/ and print' input
abcabcabc
In Perl 5, assertions provide another form of context sensitive matching.
Meanwhile, I think you're right that Perl 5 regexes cannot express all of type 2,
unless you cheat by using (?{ code }).