Re: [Monotone-devel] Re: Bug in CRLF conversions

[Yury Polyanskiy]

On Sun, 2006-01-29 at 21:32 +0100, Richard Levitte - VMS Whacker wrote:

> Nope.  Consider a binary file, which might contain anything and let's
> consider a get_linesep_conv that always returns {"LF","CR"} (because
> we run it on Mac, that's why).  Let's use a hypothetical monotone that

Well I wasn't thinking about LF<->CR case. For LF<->CRLF or CR<->CRLF
this works. 

Some motivation why maybe it still should be done. Why we want line
endings translation at all? Because frigging windows software doesn't
handle lone \n's or \r's very well (notepad, etc). Mac software I
believe is much better in this respect so they can live w/o translation.
Can anyone using Mac comment on this?

And in any case currently we have "surprising" conversion for all cases
and I suggest that we have "surprises" only in (super rare?) LF<->CR
case. Not to say that non-surprising implementation 'd be much faster.

Cheers,
 -up