Errr, since you were talking about STOR I should have said the same applies on the server side. FTP servers only convert uploaded files to the local OS format by convention. All the spec says is that the ASCII transfer itself must use CRLF's.
Anyway, something isn't right because I remember seeing problems with the line enders using a client that was known to behave correctly. Clint On 5/4/07 2:57 PM, "Clinton Foster" <[EMAIL PROTECTED]> wrote: > I think this problem may have been around for quite some time. If there > isn¹t a JIRA issue, it seems like we need to create one. > > I¹ve noticed there is sometimes confusion about ASCII file transfers because > the RFC 959 specification says ASCII mode data must always be transferred > using CRLF¹s as the line enders (regardless of the host operating system of > the client or server). But people are accustomed to thinking in terms of the > end product at the client always having line enders appropriate for the > local platform. But that conversion is entirely a function of the client > implementation. The specification says nothing about it. In other words, it > is simply by convention that most clients automatically convert files that > were transferred in ASCII mode to the local OS line enders when they persist > them (e.g. LF for UNIX, CRLF for Windows) even though the actual transfer > must be done with CRLF¹s. > > Clint > > > On 5/4/07 2:39 PM, "Peter Kostouros" <[EMAIL PROTECTED]> wrote: > >> Hi >> >> What is the expected STOR of files that use CR+LF as line separators? In >> ASCII mode I see files transferred with additional CR characters, i.e., >> CR LF -> CR CR LF. Is this correct? (Ref. IoDataConnection::transfer(...)) > >
