bug-texinfo
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [bug #38795] texi2any makes CR in output when input is mixed CR-LF a


From: Vincent Belaïche
Subject: Re: [bug #38795] texi2any makes CR in output when input is mixed CR-LF and LF files
Date: Tue, 20 Aug 2013 23:53:32 +0200
User-agent: Thunderbird 2.0.0.24 (Windows/20100228)

Karl Berry a écrit :
Vincent,

Finally getting back to this mail.

    Date: Sun, 21 Apr 2013 19:02:39 +0200
    => install-info (GNU texinfo) 4.11
    => Copyright (C) 2007 Free Software Foundation, Inc.

That's pretty old.

As far as I can see, both the current install-info (maybe I eventually
installed your patch, I don't remember) and the current makeinfo (5.1,
also 5.0 I'm sure) handle CRLF as normal line endings.  (TeX has done so
for decades.)  I tested this by making a file with CRLF line endings and
processing it on Unix.

So I removed the paragraph you pointed out from the manual.

I see from the subject line that your original file might have had mixed
CRLF and LF line endings, rather than consistently one or the other.
Inconsistent line endings are a fundamental problem and I don't promise
to support such files in all cases, although a couple tests worked out ok.

Anyway, if problems remain, please include a test file with the report.

karl

Hello Karl,

Just one clarification on he problem description: each file on its own is consistently encoded w.r.t. end of lines, but some files are using LF while some other files are using CRLF. Now, because these files intertwine together by way of including, then the problem arises.

I was not meaning that Texinfo should handle files that are not properly encoded --- that would clearly be the user's guilt --- but I came across the problem as I was pulling some files from some version control repo, and these files were created under linux, but some of them were tagged as binary (although being text) and as such not translated (then keeping the LF ending) while other were tagged as text, and as such translated.

So the problem was not even really that the person who created the files was inconsistent in handling the end of lines, but rather that the version binary/ascii flag was inconsistently set.

Given that this sort of problem may probably arise again to some people, it would be nice if Texinfo could handle it properly --- with some warning message would be even nicer to have.

VBR,
  Vincent.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]