Re: PCL-CVS: more diff options

[Kai Grossjohann <kai.grossjohann@gmx.net>]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

>> I am in the habit of doing M-x cvs-update RET in the morning.  Say that
>> I then see that the file foo has been updated.  Now what I routinely want
>> to know is what has changed by this update.
>
>> WIBNI PCL-CVS offered a convenient way to see this?
>
> The problem is that CVS doesn't really offer any way to do that.
> If the file was locally modified, you can use the most recent
> .#<file>.<rev>.

Well, just diffing the most recent two revisions would be fine, I
think.

Hm.

Oh!  What if the file has changed by more than one revision while I
wasn't looking?  Yes, indeed.  Hm.

> But otherwise, CVS doesn't remember which files were updated and
> neither does it remember what was the previous revision number.

Hm, yes.  Hm.  It should ;-)

> PCL-CVS could try to do that, but for it to work even if you
> exit/rerun PCL-CVS, we'd need to store it (in CVS/<something>
> typically), which I'd rather avoid if possible).

Hm.  PCL-CVS would have to do "cvs status" first, so that it knows
the state, right?  Then a "cvs update" will say what has changed and
so it can do the right diff.

But if something was stored in CVS/foo, then maybe a history of
updates would be nice.  Then CVS could look in CVS/foo for when was
the last update performed and then diff on the date.  WDYT?

> What I do instead is to use `dh' *before* updating.  It also suffers from
> problems, such as the race condition (someone might commit something
> between the `dh' and the update).

Ah, yes.  Maybe I can remember to do that in the future.

I wish people would just write ChangeLog entries like the good boys.
But ChangeLog entries don't pay the rent and so I'll have to make do
without them, I guess.  (I am running cvs2cl from time to time, tho.)