Why did every line in w32fns.c get a ^M?

Why did every line in w32fns.c get a ^M?

[Lennart Borgman]

When I just updated I got a merge conflict in w32fns.c. At the same
time all the lines in that file got a ^M at the end.

Why did they get that? (This is on w32.)

Re: Why did every line in w32fns.c get a ^M?

[Juanma Barranquero]

On Thu, Jul 1, 2010 at 12:37, Lennart Borgman <lennart.borgman@gmail.com> wrote:

> When I just updated I got a merge conflict in w32fns.c. At the same
> time all the lines in that file got a ^M at the end.
>
> Why did they get that? (This is on w32.)

Did you have local changes on w32fns.c? Perhaps you mixed CRLF vs LF.
Have you tried reverting the file?

    Juanma

Re: Why did every line in w32fns.c get a ^M?

[Lennart Borgman]

On Thu, Jul 1, 2010 at 12:42 PM, Juanma Barranquero <lekktu@gmail.com> wrote:
> On Thu, Jul 1, 2010 at 12:37, Lennart Borgman <lennart.borgman@gmail.com> wrote:
>
>> When I just updated I got a merge conflict in w32fns.c. At the same
>> time all the lines in that file got a ^M at the end.
>>
>> Why did they get that? (This is on w32.)
>
> Did you have local changes on w32fns.c? Perhaps you mixed CRLF vs LF.
> Have you tried reverting the file?


I have a lot of changes there, but I don't think there were any mix of
CRLF vs LF.

Re: Why did every line in w32fns.c get a ^M?

[Juanma Barranquero]

On Thu, Jul 1, 2010 at 12:56, Lennart Borgman <lennart.borgman@gmail.com> wrote:

> I have a lot of changes there, but I don't think there were any mix of
> CRLF vs LF.

The sudden appearance of ^M at the end of lines is an almost-100% sure
sign of such a mixup. When we're talking of a file that in a Bzr
checkout is LF, but it is often edited with CRLF-prone Windows tools,
the likelihood surpasses 100% :-)

So, please save your w32fns.c, revert it, and if it works, you know
the problem is in your changes... Certainly my w32fns.c has no trouble
whatsoever right now.

    Juanma

Re: Why did every line in w32fns.c get a ^M?

[Lennart Borgman]

On Thu, Jul 1, 2010 at 1:00 PM, Juanma Barranquero <lekktu@gmail.com> wrote:
> On Thu, Jul 1, 2010 at 12:56, Lennart Borgman <lennart.borgman@gmail.com> wrote:
>
>> I have a lot of changes there, but I don't think there were any mix of
>> CRLF vs LF.
>
> The sudden appearance of ^M at the end of lines is an almost-100% sure
> sign of such a mixup. When we're talking of a file that in a Bzr
> checkout is LF, but it is often edited with CRLF-prone Windows tools,
> the likelihood surpasses 100% :-)

The only tool I used editing it is Emacs ;-)

But it looks like w32fns.c somehow got saved with CRLF line endings (I
have no idea why) and it looks like bzr merge could not handle that at
all.

> So, please save your w32fns.c, revert it, and if it works, you know
> the problem is in your changes... Certainly my w32fns.c has no trouble
> whatsoever right now.
>
>    Juanma
>

Re: Why did every line in w32fns.c get a ^M?

[Juanma Barranquero]

On Thu, Jul 1, 2010 at 13:36, Lennart Borgman <lennart.borgman@gmail.com> wrote:

> But it looks like w32fns.c somehow got saved with CRLF line endings (I
> have no idea why)

I don't think it was Bazaar, so you're the likely culprit somehow... :-)

> and it looks like bzr merge could not handle that at all.

How could it? If the local copy suddenly has CRLF in every line, *all*
lines have changed and it has no anchors to detect where to apply
incoming changes.

    Juanma

Re: Why did every line in w32fns.c get a ^M?

[Lennart Borgman]

On Thu, Jul 1, 2010 at 1:46 PM, Juanma Barranquero <lekktu@gmail.com> wrote:
>
>> and it looks like bzr merge could not handle that at all.
>
> How could it? If the local copy suddenly has CRLF in every line, *all*
> lines have changed and it has no anchors to detect where to apply
> incoming changes.

You mean bzr can't be smart enough to recognize different line
endings? Of course not if it considers everything as binary, but that
is not necessary.

Re: Why did every line in w32fns.c get a ^M?

[Eli Zaretskii]

> From: Lennart Borgman <lennart.borgman@gmail.com>
> Date: Thu, 1 Jul 2010 14:56:31 +0200
> Cc: Emacs-Devel devel <emacs-devel@gnu.org>
> 
> On Thu, Jul 1, 2010 at 1:46 PM, Juanma Barranquero <lekktu@gmail.com> wrote:
> >
> >> and it looks like bzr merge could not handle that at all.
> >
> > How could it? If the local copy suddenly has CRLF in every line, *all*
> > lines have changed and it has no anchors to detect where to apply
> > incoming changes.
> 
> You mean bzr can't be smart enough to recognize different line
> endings? Of course not if it considers everything as binary, but that
> is not necessary.

I actually think the default Bazaar operation regarding EOLs is very
nice.  Even in your case, it did help you identify the problem right
away, didn't it?

But if you don't like the default and want to get yourself into more
trouble, read the section "End of Line Conversion" in the Bazaar User
Reference manual (in the "Concepts" chapter).

Re: Why did every line in w32fns.c get a ^M?

[Lennart Borgman]

On Thu, Jul 1, 2010 at 7:03 PM, Eli Zaretskii <eliz@gnu.org> wrote:
>> From: Lennart Borgman <lennart.borgman@gmail.com>
>> Date: Thu, 1 Jul 2010 14:56:31 +0200
>> Cc: Emacs-Devel devel <emacs-devel@gnu.org>
>>
>> On Thu, Jul 1, 2010 at 1:46 PM, Juanma Barranquero <lekktu@gmail.com> wrote:
>> >
>> >> and it looks like bzr merge could not handle that at all.
>> >
>> > How could it? If the local copy suddenly has CRLF in every line, *all*
>> > lines have changed and it has no anchors to detect where to apply
>> > incoming changes.
>>
>> You mean bzr can't be smart enough to recognize different line
>> endings? Of course not if it considers everything as binary, but that
>> is not necessary.
>
> I actually think the default Bazaar operation regarding EOLs is very
> nice.  Even in your case, it did help you identify the problem right
> away, didn't it?


I think it is rather stupid actually.

It could have warned me of the situation instead.


> But if you don't like the default and want to get yourself into more
> trouble, read the section "End of Line Conversion" in the Bazaar User
> Reference manual (in the "Concepts" chapter).


Thanks.

Re: Why did every line in w32fns.c get a ^M?

[Juanma Barranquero]

On Thu, Jul 1, 2010 at 19:03, Eli Zaretskii <eliz@gnu.org> wrote:

> Even in your case, it did help you identify the problem right
> away, didn't it?

Well, no, he first suspected of some general trouble afecting all users... ;-)

    Juanma