Re: -v ?

Craig A Summerhill wrote;

> > If you have one of the environment variables set, I think it is used
> > although it may not be shown uncommented in the -v output (since it
> > wasn't used in the actual config file).
> >
> > Should I make it appear uncommented in such a case?
>
> Personally, I think this makes sense. I *assumed* the output from the
> -v flag was telling me the settings in place (especially when used in
> combination with the -c flag to read an external .hmrc config file).

Yes, I think you're right. I just never use any environment variables myself, that's why I tend to forget about them :-)

> > 1. Internal defaults
> > 2. Environment variables
> > 3. Config file items
> > 4. Command line options

> If hypermail is using a setting from one of the four methods above it
> just make sense to me that it would appear uncommented in the -v output.
> Thus you can use the -v flag to output a .hmrc file which can be
> edited to your liking...

I take it you don't mean the (1) ? I think it is useful to not display values that are set to internal defaults as uncommented, as they can then be changed in future versions more easily.

> I have to say, I have been confused since I first started using Kevin's
> original 1.x hypermail in trying to determine which settings were
> being enforced when.

Yes, me too. I've been struggling to sort all that out.

> Defaults behaviors used to be compiled in by
> editing the options.h file. But I have opted to put as much as
> possible in external .hmrc config files because I found I was getting
> totally unpredictable behaviors from hypermail on various executions
> otherwise.

Yes. That also makes hypermail a lot easier to debug.

> The down side of this is that I need to maintain a config
> file for each separate archive I want to markup -- which can be time
> consuming. But it works.

Yes. Although you could actually make a global config file and then only add the local changes for each separate package. Perhaps the number of local changes are even few enough to be specified on the command line.

> But I am still not clear on what hypermail
> is doing to set default behaviors for various settings as per (2) in
> your list above.

When it goes through the list of all variables, it checks for the environment variables with the same names. If the environment variables are set, those values are used. Otherwise the built-in values are used.

> Some default behaviors must be "compiled in" to hypermail as per (1) in
> your list above.

Yes, those values are set in the source and is not meant to be locally modified.

> If defaults are not compiled into the binary

All variables have built-in default values.

> and they
> are not specified as per (3) in your list above, from where are they read?

1 is always present. 2, 3 and 4 are optional.

> Are you talking about shell environment settings?

When I say "Environment variables" I mean standard shell environment variables, yes.

 / Daniel Received on Wed 14 Apr 1999 10:24:56 AM GMT