[darcs-users] high-level UTF-8 feedback

Trent W. Buck twb at cybersource.com.au
Mon Jan 4 09:06:06 UTC 2010


Reinier Lamers <tux_rocker at reinier.de> writes:

>  * Investigate how to handle the ICU dependency on Windows; if we have
>    to, just bundle the ICU source.

>From the Debian integrator to the release manager:

It will annoy me if the source tarball for 2.4 contains icu, because
I'll have to either roll a new tarball (i.e. unbundle it) or document
the copyright details of all of icu.

Since I don't care about Windows users, I'd prefer to provide them with
pre-built binaries rather than include icu in the source tarball.

> Decoding according to locale is done by Haskeline, which I thought
> could deal with input that is malformed according to the current
> locale. I can't find this in the Haskeline documentation anymore
> though. Judah, can you confirm that it is really failsafe?

Isn't it trivial to test this, by doing e.g.

    darcs init
    touch x
    LC_ALL=C darcs rec -la x

...and pasting an α or א into the haskeline prompt?  Presumably if it's
validating input, it'll bork on those characters since they'll have at
least one high-bit byte.



More information about the darcs-users mailing list