Re: ci breaks links unnecessarily & when it says file is unchanged

L A Walsh wrote:
> Paul Eggert wrote:
>> L A Walsh wrote:
>>> 1) Why are hardlinks _ever_ broken on checking a file in?
>>
>> ci has always done that, since it may need to change the file and/or make it
>> read-only.
> ---
>    Just because something had a design flaw to begin with is not a justification
> for it remaining a bug.  If it *needs* to make it read-only, or needs to make a
> change, then it is expected, but just to break links "in case".

I didn't say anything about "in case". You asked why are hardlinks _ever_ 
broken. They need to be broken when the working file's data or metadata need to 
change. I agree they don't need to be broken when nothing at all changes.

>>> 2) especially in in the case when it says "file is unchanged"
>>
>> The diagnostic could be improved, yes. "file is unchanged" means the revision
>> is unchanged, not that the working file is untouched.
> Yet it also indicates the original intent at the same time.  If nothing needs to
> be done

No, sometimes the working file needs to be changed even if there is no 
difference in what's checked in, due to $Id$ expansion and the like.