RE: using the date relationship of two older files as a prerequisite

[Cook, Malcolm]

would it serve your purpose to set the time on T in the rule that creates it to 
be the time on O (instead of the current time)


~Malcolm


> -----Original Message-----
> From: address@hidden [mailto:address@hidden On Behalf Of Brian
> J. Murrell
> Sent: Friday, December 07, 2012 10:42 AM
> To: address@hidden
> Subject: Re: using the date relationship of two older files as a prerequisite
> 
> On 12-12-07 10:53 AM, Mason wrote:
> >
> > OK, so "O" is just like a normal source file; it is not created by make,
> > and it is created/updated by your text editor.
> 
> Correct.  "O" is not created by make, but it's not created in real-time
> either like an editor file is.  That is, it can get updated but even
> when it does while it will be newer than it was before, it will still be
> older than the target "T".
> 
> > This paragraph does not make sense (to me).
> >
> > At some point, "O" is modified (possibly several times).
> 
> Correct.
> 
> > Suppose the "O" was last modified at T1.
> > Then you call make, which builds "T" at T2.
> 
> Right.
> 
> > Obviously T1 < T2
> 
> Correct.
> 
> > If "O" is modified (possibly several times) at T3,
> > then "O" is obviously newer than "T" since T2 < T3.
> 
> No.  O is modified at let's call it T1.5. (not T3).  It's newer than T1
> but older than T2.
> 
> > What am I missing?
> 
> I think you were missing that O is modified at T1.5 after T was rebuilt
> at T2.
> 
> The reason O can have a time of T1.5 after T2 is because O lives in an
> external source that make can see.  That source can be updated but the
> latency between it being updated and made available to make is so long
> that the updated O is still some time in the past, before T2 in most cases.
> 
> b.
>