[Unison-hackers] Problem with fractional mtime? ("destination updated during synchronization")

Benjamin Pierce bcpierce at cis.upenn.edu
Sun Nov 23 08:44:54 EST 2008


There is some rather tricky code in props.ml (in the sub-module Time)  
that compares file modtimes while trying to take into account some  
possible reasons why unchanged files may appear to have different  
modtimes (e.g., daylight savings time, or one filesystem being FAT,  
which can only store modtimes with a resolution of two seconds).  If  
the problem you're seeing is really with fractional modtimes, it may  
be possible to fix it by changing something here...

Best,

     - Benjamin


On Nov 23, 2008, at 7:42 AM, Hans Meine wrote:

> On Sonntag 23 November 2008, Hans Meine wrote:
>> In the meantime, I have confirmed that the same bug still exists in
>> 2.27.57.
>
> Now, I have to say that I tried to find a fix myself, but Ocaml is  
> harder to
> read than I thought.  I guess I have to wait for someone else to  
> acknowledge
> this bug and fix it.
>
>> The "rounding" in my last post does not make sense, probably it's  
>> more like
>> this:
>>
>> - int remembered_mtime = trunc(remote_mtime)
>> - ... [synchronize]
>> - check whether remote_mtime > remembered_mtime:
>>  ALARM! [difference 0.0 < x < 1.0 seconds]
>
> I have read that Ocaml does not do implicit casting (good!), and I  
> did not
> find any casts (at least not in update.ml), so maybe I am still off  
> with the
> above.
>
> It is also quite strange that it only with these sip-generated  
> header files.
> Obviously, SIP somehow triggers the fractional mtime, while other  
> programs do
> not seem to do that.  Obscure.
>
> -- 
> Ciao, /  /                                                    .o.
>     /--/                                                     ..o
>    /  / ANS                                                  ooo
> _______________________________________________
> Unison-hackers mailing list
> Unison-hackers at lists.seas.upenn.edu
> http://lists.seas.upenn.edu/mailman/listinfo/unison-hackers



More information about the Unison-hackers mailing list