On 03/01/2016 11:57 AM, Paul Eggert wrote:
On 03/01/2016 02:37 AM, Ian Abbott wrote:
Some of those patches look a little odd when viewed in an MUA anyway (Thunderbird or Icedove in my case), as the MIME charset is declared as windows-1252, but the actual text is UTF-8. (That's most prevalent in the changes to zone1970.tab by PATCH 10/10.)
Unfortunately some of the patches were still messed up, at least as received here. Patch 04 was double-encoded: it said "Content-Type: text/plain; charset=UTF-8" and "Content-Transfer-Encoding: quoted-printable" and the body contained strings like "Tucum=C3=83=C2=A1n", which decode to "Tucumán" instead of to the correct "Tucumán". As Ian notes, patch 10 was double-encoded in a different way: it said "Content-Type: text/plain; charset=windows-1252" and "Content-Transfer-Encoding: quoted-printable" and contained strings like "Le=C3=B3n", which decode to "León" instead of to the correct "León".
I am sorry Paul, for the extra work I caused you. My system is deeply configured (down to kernel level) to not use multi byte characters. This is the first project I've contributed to that it mattered. I expect that I will bite the bullet and switch to utf8 on my next complete OS upgrade.
I hope I fixed the decoding errors correctly before installing this stuff into the experimental version on GitHub.
If it is any indication, after iconv-ing them they display correctly on my system.