Charles Plessy <ple...@debian.org> writes: > First, as a sidenote, no field specifies that it may not span multiple > lines. I therefore agree with you that it is an implicit default case, > and propose to make it explicit in § 5.1 (see below).
Agreed. > I then looked at which field description specifies that they ‘may span > multiple lines’. These are the relationship fields (Depends etc., §7.1), > the Binary field, and the Uploaders field, but only in source package > control files. The Files and Checksums-* fields, on the other hand are > described as ‘multiline fields’. Lastly, nothing is specified for the > Description and Changes fields, perhaps because it is so obvious. > It looks like ‘may span multiple lines’ means that continuation lines > are allowed but newlines are not significant, and ‘multiline fields’ > means that continuation lines are allowed and newlines are > significant. At this point, I am not sure what is expected from the > parsers: deliver the value of the fields that ‘may span multiple lines’ > with or without newlines? In any casee, I find this similarity of > terminology very confusing. I therefore propose to replace ‘may span > multiple lines’ by ‘continuation lines are allowed’ and add it where it > was implicit, and add ‘continuation lines are allowed and newlines are > significant’ where needed as well. Agreed. This is very confusing right now, and I think your proposed method of rephrasing is a good one. Although another approach may be to introduce the folding terminology from RFC 5322. The distinction really is that some fields can be folded (Build-*, for example) and some fields are multi-line (Description, Files). The multi-line fields are not folded in the RFC 5322 sense, since you cannot just remove the newlines and have semantically the same content. Those fields (Description, Files) are a separate type of field that RFC 5322 doesn't have. > I noted another confusing sentence on the subject, in §5.2: ‘Many fields > are permitted to span multiple lines in <file>debian/control</file> but > not in any other control file’. Actually, I found this to be true only > for the Uploaders field. I believe this is true of all binary relationship fields and all build relationship fields as well. The dpkg-dev tools unfold all of those fields when generating *.dsc, *.changes, and DEBIAN/control files, and parers of those generated files do not have to cope with folded fields (and I believe are known *not* to be able to cope with folded fields in some cases). We should say that explicitly. > I propose to replace it by ‘Continuation lines can be permitted for some > fields in <file>debian/control</file> but not in any other control > file.’ Sounds okay to me with the above note about possibly wanting to use folding instead. -- Russ Allbery (r...@debian.org) <http://www.eyrie.org/~eagle/> -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
