Re: [Monotone-devel] Re: Looking at the code affected in bug 9752 leaves a weird taste...

[Richard Levitte - VMS Whacker]

In message <address@hidden> on Wed, 18 Aug 2004 10:46:10 -0600, Derek Scherger 
<address@hidden> said:

derek> I dug a bit further into this one, extracting, decoding and
derek> uzipping the raw file data from the db and they do seem to have
derek> doubled up LF's. I don't know what I my get_linesep_conv hook
derek> was set to when those files were created and to complicate
derek> things I'm running monotone under cygwin, whether that has any
derek> implications or not I don't know.

What actually happens is that somewhere, split_into_lines() getc
called.  That function uses a boost::tokenizer with the separator
functor boost::char_separator.  This is used in a manner that much
reminds of a series of calls to strtok_r() like this:

        char *tmp = NULL;
        char *token;
        while((token = strtok_r(input, "\r\n", &tmp))) {
                input = NULL;
                add_to_token_list(token);
        }

Basically, it's regarding \r and \n as separate token (line)
separators.  This means that the string "foo\r\n" will end up being
split in the 'lines' "foo" and "" by split_into_lines(), REGARDLESS of
what you wrote in get_linesep_conv, since split_into_lines() doesn't
take any such data.

-----
Please consider sponsoring my work on free software.
See http://www.free.lp.se/sponsoring.html for details.

-- 
Richard Levitte                         address@hidden
                                        http://richard.levitte.org/