Re: [bug-gnu-libiconv] [PATCH] Add a few useful aliases

[Bruno Haible]

Hello,

Daniel Richard G. wrote:
> The attached patch adds a few encoding-name aliases supported by the glibc 
> and Solaris implementations of iconv(1), and uncomments the "UTF8" alias. 

Aliases should be reduced to a minimum. Adding more aliases causes more
software to appear broken.

IANA maintains an official list of encodings and aliases:
http://www.iana.org/assignments/character-sets

This is how these aliases spread: A package, in this case glibc, starts to
allow people to write "UTF8" instead of "UTF-8". Then people get used to it,
ask for the next piece of software to support that alias. And so on. It
spreads like a chain letter, and has no real benefit: people can just as
well use the standardized encoding names.

To make it concrete:
There is an encoding UTF-32LE. Google hits: 25.300.
Some people spell it incorrectly: UTF32LE. Google hits: 3000.
Other people misplace the dash: UTF32-LE. Google hits: 673.

Google also shows that the string UTF-32LE is used in code 8000 times. So
if you ask for all packages to support aliases for this encoding, up to
8000 pieces of code need to change. This is a stupid request. Fix the
spelling at the origin.

> These have been sorely missed by my employer, who takes the view that the 
> hyphens in the canonical names are redundant.

Then all software would also have to support U-T-F8 and similar names. No!
It's pointless.

Bruno