AutoGen: The Automated Program Generator

Looks good, sir. Thank you.

So, my apologies for all the noise. It's been a long time since I looked
at the mod time stuff. By default, --source-time _is_ disabled. It
wasn't
for --no-definitions by mistake (bug). So, setting the time where you
did
needs to be done, but conditionally:
if (! ENABLED_OPT( SOURCE_TIME ))
outTime = time( NULL );
and the other stuff discovered needs fixing, too. All done. 5.9.10pre9
I'm done playing today. Cheers - Bruce

For your purposes, consider using: --no-source-time
or: export AUTOGEN_SOURCE_TIME=no
or: echo no-source-time >> ~/.autogenrc

I definitely agree about the configurable - something that will allow an
Autogen user to always get the current time as the output file mtimes, no
matter the mtimes of the definitions file, templates, extraction sources,
or Scheme library in use. Do you want me to file this as a feature request?

Yep. The extraction stuff has an effect, too. The correct thing is using
some sort
of file fingerprint and not using time stamps at all. That's out of
scope, though.
Here, my intention is to remove the move-if-change problem that typically
gets
solved with a script by that name. Setting the result modification time
needs to
be some sort of configurable.....Thanks

Thanks for the new prerelease, Bruce.

I saw that you had a comment in the defLoad code about how the output time
stamp should be tracking the most recently modified template, and I did
consider going down that route. But, I ended up talking myself out of it
for a couple of reasons:
(1) It does not account for files which are 'extract' sources. They may be
more recent than the template, and if the extraction sources are deps in
your make rule for the output file, then the rule will never be satisfied.
(2) I couldn't think of any really good reason why one second after the
most recently modified template was better than the current time - other
than that it is consistent with how the definitions mtime is handled. The
typical behavior I expect from other programs is that the file modification
time is the time when the file was modified. ;-)

So, I don't have any strong objection to what you did in tpLoad.c, but I
think extraction sources might now need to be brought into consideration.

Just my two cents....

I think, though, the correct patch is to ensure that the output
time stamp is always at least as recent as any template that's
been opened. There must always be at least one...
Please look here for the 10pre8 (or later) version:
http://autogen.sourceforge.net/data/

Oops. I've rarely used it and I suspect not too many others do, either.
Thank you! 5.9.10 will be out shortly.