This seems to be related to http://savannah.gnu.org/bugs/?54857
‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Thursday, February 14, 2019 2:49 PM, <[email protected]> wrote: > It seems like LibreJS, at least in Firefox, can change the expected > default character coding, which I think is a bug. Here's how to > reproduce: > > In Firefox with disabled LibreJS I go to > https://cl.lingfil.uu.se/~starback/char.html . That test only > contains > > ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- > > <!DOCTYPE html> > <html> > <body> > > Lätiñ 1. > > > </body> > </html> > > coded in Latin-1. The web server doesn't say anything about character > encoding (as confirmed by wget -S) and it is shown by Firefox as the > expected "Lätiñ 1.". > > Then I activate LibreJS and reload the page. Now it is shown as > "L�ti� 1." instead, and I notice that at "View→Text Encoding" now > "Unicode" is ticked instead of "Western" as it was before. > > It is surely good practice to always specify character encoding on > the web, but the HTTP standard still specifies that iso-8859-1 is the > default, so I think this should work, and LibreJS introduces a bug. > > (Versions: Firefox 60.5.0esr for CentOS on GNU/Linux; > GNU LibreJS 7.19rc3.)
