On 02/11/2007, Ricardo Mones wrote: > You cannot pretend claws-mail (or any other program) to know how to > decode arbitrary encoded data from all its potential callers.
Ricardo, this is the code that gets called when claws is invoked with 'claws --compose mailto:something-or-other'. Since everything from the mailto: onwards is by definition an URI, it *must* be URI-encoded (it would be a bug if it weren't!). So it's not a matter of "decoding arbitrary encoded data", it's about handlilng URIs the way they are supposed to be handled. > The data should reach command line already decoded by the browser > calling claws-mail. This is incorrect. Supose you have a link in a web page to send a message to "J. Random Hacker & Co. <[EMAIL PROTECTED]>". We could not send this address to claws-mail with claws-mail --compose mailto:J. Random Hacker & Co. <[EMAIL PROTECTED]> For two reasons: 1) the shell will separate the URL after "J." 2) even if we quoted it to make the whole address one word, it results in an invalid URI, because it contains naked spaces, punctuation and even an "&" which would get the URI parser mightily confused > If not that's a bug in the browser, not in claws-mail. Which browser are you > using? I'm using a script I wrote (attached) to make it possible to use claws as a MUA for debian's reportbug (in fact, this very bug report was sent using it!). To use it with reportbug, you use reportbug --mua /path/to/claws-recompose package Be kind to it, it's still unpolished, I intend to contribute it to claws tools. But you can reproduce it just as easily with mozilla-* (and I guess with any browser) with the attached trivial html file. This is definitely a bug in claws, not anywhere else. Fede
claws-recompose
Description: Binary data
signature.asc
Description: PGP signature
