I'm curious why Evolution uses an ass-backwards internal mapping of mail folders to files. Folders are not mapped directly to directories; no, a folder name is a prefix for several file names with predefined suffixes, one of them is a directory for subfolders. I didn't try but I guess, God forbid you from creating a folder with one of these suffixes in the name. Why it couldn't be made the natural way (a folder is mapped to a directory containing predefined files for contents/indexes, subdirectories are mapped to subfolders), is completely beyond me.
Come on, writing routines that grow or shrink a chunk of memory, or search that memory as a series of key-> value pairs, doesn't require any special libraries.
Come on, writing routines that have already been written, stepping into bugs and inefficiencies that have already been taken care of elsewhere, is totally unnecessary in this day and age. This is what glib is for.
These "threads" from the pretty picture are not what I'm talking about. They are software controlled, which means there must be some code that takes care of them at once.
Yeah, yeah, just imagine a Beowulf cluster of... nevermind:) But what prevents all these programs from stepping on each others' toes when they submit tasklets to SPUs? Will the arbitration be performed benevolently by a mutual convention or enforced by the OS?
But how all this is going to be multitasked? When my process is being switched out in the main CPU, should the running SPUs be also suspended somehow and their context saved along with the main context? Since their local memory isn't protected in any way, that would be quite a massive context, wouldn't it? If this is not to be done, access to the SPUs should be policed by the OS. Say, while some process has a device opened that controls access to an SPU, no other process can open the same device.
The article also points out that the SP floats aren't truly 754-compliant, as they round-toward-zero on cast to int.
As far as I remember from implementing the spec years ago, the rounding mode can be varied. Indeed there are C runtime functions on many platforms that set this and other properties for floating point operations.
We could probably write an amazingly fast MP3 encoder, but if it's only single precision floating point, then maybe we won't.
Why? Single precision floating point can accomodate up to 23 bits of precision, and full 24 if you consider the sign (all sound applications should use zero-centered FP samples because floating point becomes more precise towards zero). Sure many modern digital sound systems are exactly 24 bit so there is no margin for errors, but the lowest bits are for marketing and bit-padding purposes anyway.
Wait, so it's just as hard to parse, has the same expressive power, and certain classes of errors are *harder* to track down.
It doesn't even have a 1-to-1 mapping, contrary to what the grandparent states, at least not immediately. XML has attributes with enforced map-like semantics along with children that can represent lists. And there is a standard and mandatory way to specify character encoding for an XML document. S-expressions seem to be always specified in some externally-imposed character set.
There are many aspects of what constitutes a good programming language for different people:
Conceptually perfect and aesthetically beautiful.
Effective at making the computer do stuff you want it to.
Facilitates development of large projects by ever-changing teams of middle-to-good programmers.
Has decent safety/security model.
...
It seems that he values only the first of these qualities. So what if your favorite language can describe an interpreter for itself. Can it also describe system calls and threading primitives? Programs still have to run on silicon CPUs in operating system environments. There are no processors available that could grind S-expressions natively. It takes some translation to do that. If you know it might be otherwise, why not step down from your high horse and give some insights, instead of making witty remarks about the unwashed masses?
Late binding and dynamic typing are good if you are doing little programs, or you've got that air of impeccability around you. Us mere mortals would rather thank the compiler for finding stupid mistakes at the translation time.
Yes, admit it: the world is full of stupid people, and it is run by them. If there is a tool that fits the job, even if it doesn't require a science degree to operate, it shouldn't be frowned upon.
Reading this reminded me about that portion of the standard C++ library which is all about operations on vector data. So, my question is: could an std::valarray specialization for processor-supported types serve as a basis for portable SIMD support in C++?
Moto has implemented what they call haptics in their E398 phone for that extra youth-culture oomph. Hold it while it rings/plays and you'll feel that the oomph is quite literal.
Back when I was a teen, me and my father seen something once in the night sky, in nothern Russia. It looked like a pale greenish spot that moved at a constant speed for a few seconds. Dad said it must be some big-ass (OTH?) radar that uses the excited ionosphere for a mirror. Later I read the stories about the Russian Woodpecker and thought dad might be right.
It probably won't. The traffic is distributed around Skype's P2P network. So, the corporate abuse will be sustained by, you guessed it, the abusive corporate users.
and then Kollontai said he was like totally quitting cos no-one respected his L33T SKILLZ! Wasn't Kollontai a 'she'?
And in the end, we can see that both projects failed miserably, the longer lasting just took 70 more years. However, if we consider Chinese Communism(TM) a fork, it's still alive and kicking, even if it has had to undergo considerable, erm, refactoring.
there's no good reason to state the name of the element twice (at beginning and end)
There is. Any possible screwup is caught immediately where it occurs, not 10 closing tags and 50 sibling subtrees later with the diagnostic of "oh, there's some tag mismatch in all this mess".
Attributes are pointless, basically just another way to do sub-elements.
Attributes reflect a slightly different metaphor than that of children: they are unique properties with simple text values. They cannot occur repeatedly in an element as children may be allowed to. Also, there are many markup applications of XML (XHTML, for instance) where you can strip all XML tags and come up with reasonably readable text. Attributes conceal stuff that is not supposed to be in that stripped text. Of course this means nothing to clueless DTD/schema designers who throw everything in an element soup and then complain about XML being hard to work with.
The comments seem to be full of contempt for teenage -angst inane ramblings that are common on LJ. Come on. It's not like you are forced to read through this stuff. I have a few "friends" there at LJ, some of them net.celebs, and I like their posts. It's the matter of whose writings do you find interesting, and you are free to be completely unaware of the rest. Why all the vitriol?
My first thoughts when I saw the headline in my RSS bookmark were: "Psychic Wars? Wikipedia? Boohoo!" Boy was I disappointed when I read what the entry was about.
Slashdot Mirror
I'm curious why Evolution uses an ass-backwards internal mapping of mail folders to files.
Folders are not mapped directly to directories; no, a folder name is a prefix for several file names with predefined suffixes, one of them is a directory for subfolders. I didn't try but I guess, God forbid you from creating a folder with one of these suffixes in the name.
Why it couldn't be made the natural way (a folder is mapped to a directory containing predefined files for contents/indexes, subdirectories are mapped to subfolders), is completely beyond me.
Come on, writing routines that grow or shrink a chunk of memory, or search that memory as a series of key-> value pairs, doesn't require any special libraries.
Come on, writing routines that have already been written, stepping into bugs and inefficiencies that have already been taken care of elsewhere, is totally unnecessary in this day and age. This is what glib is for.
You were rather quick to jump ship, because the said prospect never materialized and is unlikely to materialize any time soon.
These "threads" from the pretty picture are not what I'm talking about. They are software controlled, which means there must be some code that takes care of them at once.
Yeah, yeah, just imagine a Beowulf cluster of... nevermind :)
But what prevents all these programs from stepping on each others' toes when they submit tasklets to SPUs? Will the arbitration be performed benevolently by a mutual convention or enforced by the OS?
But how all this is going to be multitasked?
When my process is being switched out in the main CPU, should the running SPUs be also suspended somehow and their context saved along with the main context? Since their local memory isn't protected in any way, that would be quite a massive context, wouldn't it? If this is not to be done, access to the SPUs should be policed by the OS. Say, while some process has a device opened that controls access to an SPU, no other process can open the same device.
The article also points out that the SP floats aren't truly 754-compliant, as they round-toward-zero on cast to int.
As far as I remember from implementing the spec years ago, the rounding mode can be varied. Indeed there are C runtime functions on many platforms that set this and other properties for floating point operations.
We could probably write an amazingly fast MP3 encoder, but if it's only single precision floating point, then maybe we won't.
Why? Single precision floating point can accomodate up to 23 bits of precision, and full 24 if you consider the sign (all sound applications should use zero-centered FP samples because floating point becomes more precise towards zero). Sure many modern digital sound systems are exactly 24 bit so there is no margin for errors, but the lowest bits are for marketing and bit-padding purposes anyway.
Carmack builds a large base on Mars to conduct some shady experiments, and, well, you know the rest?
Wait, so it's just as hard to parse, has the same expressive power, and certain classes of errors are *harder* to track down.
It doesn't even have a 1-to-1 mapping, contrary to what the grandparent states, at least not immediately.
XML has attributes with enforced map-like semantics along with children that can represent lists. And there is a standard and mandatory way to specify character encoding for an XML document. S-expressions seem to be always specified in some externally-imposed character set.
It seems that he values only the first of these qualities. So what if your favorite language can describe an interpreter for itself. Can it also describe system calls and threading primitives? Programs still have to run on silicon CPUs in operating system environments. There are no processors available that could grind S-expressions natively. It takes some translation to do that. If you know it might be otherwise, why not step down from your high horse and give some insights, instead of making witty remarks about the unwashed masses?
Late binding and dynamic typing are good if you are doing little programs, or you've got that air of impeccability around you. Us mere mortals would rather thank the compiler for finding stupid mistakes at the translation time.
Yes, admit it: the world is full of stupid people, and it is run by them. If there is a tool that fits the job, even if it doesn't require a science degree to operate, it shouldn't be frowned upon.
No technical intrinsic merits push XML over some other format
Name one, please.
Reading this reminded me about that portion of the standard C++ library which is all about operations on vector data. So, my question is: could an std::valarray specialization for processor-supported types serve as a basis for portable SIMD support in C++?
Moto has implemented what they call haptics in their E398 phone for that extra youth-culture oomph. Hold it while it rings/plays and you'll feel that the oomph is quite literal.
Back when I was a teen, me and my father seen something once in the night sky, in nothern Russia. It looked like a pale greenish spot that moved at a constant speed for a few seconds. Dad said it must be some big-ass (OTH?) radar that uses the excited ionosphere for a mirror.
Later I read the stories about the Russian Woodpecker and thought dad might be right.
AKA rubber hose cryptanalysis?
before Skype shuts down from corporate abuse?
It probably won't. The traffic is distributed around Skype's P2P network. So, the corporate abuse will be sustained by, you guessed it, the abusive corporate users.
I am astonished to find out that at my current job, we are trying to solve a grand challenge.
Try EverNote. It's still in beta, but it's free.
and then Kollontai said he was like totally quitting cos no-one respected his L33T SKILLZ!
Wasn't Kollontai a 'she'?
And in the end, we can see that both projects failed miserably, the longer lasting just took 70 more years. However, if we consider Chinese Communism(TM) a fork, it's still alive and kicking, even if it has had to undergo considerable, erm, refactoring.
there's no good reason to state the name of the element twice (at beginning and end)
There is. Any possible screwup is caught immediately where it occurs, not 10 closing tags and 50 sibling subtrees later with the diagnostic of "oh, there's some tag mismatch in all this mess".
Attributes are pointless, basically just another way to do sub-elements.
Attributes reflect a slightly different metaphor than that of children: they are unique properties with simple text values. They cannot occur repeatedly in an element as children may be allowed to.
Also, there are many markup applications of XML (XHTML, for instance) where you can strip all XML tags and come up with reasonably readable text. Attributes conceal stuff that is not supposed to be in that stripped text.
Of course this means nothing to clueless DTD/schema designers who throw everything in an element soup and then complain about XML being hard to work with.
This has been discovered by a colleague of mine.
Can't remember details. I might ask him next time I see him.
The comments seem to be full of contempt for teenage -angst inane ramblings that are common on LJ. Come on. It's not like you are forced to read through this stuff.
I have a few "friends" there at LJ, some of them net.celebs, and I like their posts. It's the matter of whose writings do you find interesting, and you are free to be completely unaware of the rest. Why all the vitriol?
Of course all this doesn't help against human error.
Microsoft is known to have released components with colliding UIDs.
What do they need .asia for? What about the other parts of the world? There's .eu, but that's for the European Union, a political entity.
My first thoughts when I saw the headline in my RSS bookmark were: "Psychic Wars? Wikipedia? Boohoo!"
Boy was I disappointed when I read what the entry was about.