Moved sendsbr.c into send.c as it is not compiled into other binaries anymore.
Cosmetic changes.
Better mime types for attachments. Use some command; e.g. `file --mime'. Up to now attachments had either been text/plain or application/octet-stream, if one hadn't added mhshow-suffix- profile entries for any possible type. Now we call mimetypequeryproc (default: `file -b --mime' for GNU file), this can be changed with the profile entry `Mime-Type-Query'. This command will be executed with the attachment filename as last argument. If your file(1) command does not print MIME types, you can still grep your profile for matching mhshow-suffix- entries. However, I suggest to store this data somewhere else. IMO the profile is not the right location for it.
Layouting and comment changes. (minor stuff)
forw/repl/dist annotations won't add address lists anymore. spost does not support it and I wonder if we need it. Remove the undocumented -idanno switch, actually a no-op, from spost.
Removed the undocumented -unique switches. Seems they were JLR-only stuff.
Fixed -mime for forw and repl. If the value of an attachment header starts with `+', then it is treated as the specification of a message in MH style: ``+folder msg''
Activated Jon's attachment system by default and steamlined it. I had done similar work already some time ago. See: http://www.mail-archive.com/nmh-workers@nongnu.org/msg02112.html This time it's the second go. I changed Jon's attachment system to have a default attachment header field name, or get it from the profile. I removed the command line options. The new profile name is: `Attachment-Header'. Stored in attach_fmt. I removed the the choice between several attachment formats. A message gets mimeified if it has attachment headers or (and this is new) contains any non-ASCII character. The body won't be scanned for lines matching /^#/. The MIME structure is created automatically. The content-typing of unknown attachments is still poor and needs to be improved. You shouldn't use the `mime' command at the whatnow prompt anymore.
s/vfork/fork/ I truly believe that vfork is wrong. Forking should not be too slow for us.
Relayouted all switch statements: case aligns with switch. Desired style is: switch (n) { case foo: ... }
We have `NULL' in order to not need to write `(char *)0' anymore.
Removed void casts for ignored return values of function calls. IMO, this distracts from the important without being much worth.
Completely reworked the path convertion functions Moved everything (from sbr/getfolder.c and sbr/m_maildir.c) into sbr/path.c, but actually replaced the code almost completely. See h/prototypes.h for the function changes. sbr/path.c provides explaining comments on the functions. None of them allocates memory automatically. Additionally: - Like for other ``files'', `inc -audit file' places file relative to the cwd, not relative to the mh-dir. This is for consistency. - Replaced add(foo, NULL) with getcpy(foo), which ist clearer.
Renamed r1bindex() to mhbasename(), to make its function becomes clear. All calls to r1bindex() had '/' as second argument. For everything else one can use strrchr(3) instead of r1bindex().
Removed the space between function names and the opening parenthesis.
Reformated comments and long lines
Removed the split feature of send. It would split large messages into several partial ones, using MIME. But if a message contains one large file attached, it won't solve the problem. We better go without this feature and leave it to the user.
Rearranged whitespace (and comments) in all the code! The primary goal was to only indent with tabs and not mix them with spaces. Also, no more tabs should appear past the first printable char on each line. Now editing the sources will be much easier, no matter which editor one uses, and rearranging won't break aligned code. (Beware, the changeset is huge, but the `diff -w' changes are moderate.)
Removed mhn, as it was already replaced by mhlist/mhshow/mhstore.
Remove RCS keywords, since they no longer work after git migration.