how about a meditation? ;)
Given the kneejerk responses in this thread, I think that would be pointless.
In a nutshell:
- Variable length encoding is a "party trick" that has no place in modern IT.
On disk, standard compression algorithms are far more effective.
In memory, whatever small space savings seemed desirable in 199x; makes no sense when you can get a 64GB DIMM for £140.
The penalty for any and every algorithm that slices and dices strings is huge. The inability to directly index into strings screws with every advance searching, sorting and comparison algorithm ever devised.
It is a ridiculous state of affairs that this has been foisted on the world.
- When you have a 21-bit codepoint space, it makes no sense whatsoever to jump from 16-bit encoding to 32-bit encoding.
A fixed, 3-byte/24-bit encoding would cover every eventuality. It would allow the representation of every character in indexable strings; and leave space for embedded identification!
- No headers.
If I suggested that I would distribute multiple formats of graphics file -- png/jpg/gif/tiff; 2/4/8/24/32 bpp; huffman/RLE/LZW/none; RGB/CMYK/YCbCr etc.-- in headerless files and then suggest that people could use heuristics to try guess what they are; I'd be rightly condemned as crazy.
What the Unicode Consortium has foisted on the world is exactly equivalent.
It exists; and railing against it now is pretty pointless; but I'm at a (apparently enviable) point in my career and life where I do not have to deal with this crock, so I do not.
With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
In the absence of evidence, opinion is indistinguishable from prejudice.
| [reply] |
Given the kneejerk responses in this thread, I think that would be pointless.
Okay.
I started to write a long reply, but this is indeed probably not the best place to discuss that... I understand your point.
UTF-8 is a bit lame, and yet it's the best general-purpose encoding today (at least, among those that are actually used). The "space savings" probably always was the least interesting thing about it.
| [reply] |
this is indeed probably not the best place to discuss that
We could move to email if there is more to say? (But how to exchange emailids with Anonymonk?)
and yet it's the best general-purpose encoding today
That's a bit like saying Kim Jong-un is the best leader in NK :)
For the web -- with what are effectively signatures embedded in the Accept-Charset header -- it works well enough, but the moment you put it into a file there is no provision for identifying the particular encoding, and you're back to guessing.
With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
In the absence of evidence, opinion is indistinguishable from prejudice.
| [reply] |