Re: [PATCH] Implement new handling for margin and line-width settings.

[Carl Sorensen]

On 8/12/09 11:15 AM, "Michael Käppler" <address@hidden> wrote:

> Hi Carl,
>> Usually we like to keep responses to reviews on the -devel list, so I've
>> copied this there.
>>  
> Sorry, that wasn't intended. It was late yesterday and I didn't notice I
> pressed the wrong button... ;)

Glad I'm not the only one who does this.

>> Oh -- I misunderstood the meaning of the FIXME.
>> 
>> I think I'd recommend changing the settings to something that is rational.
>> Since we're using the changed settings, we might as well change them so we
>> don't keep getting errors.
>>  
> Am I right you suggest to modify the output-def just once but permanent?

I meant to change the paper-def so that it would be consistent for the rest
of the processing (it would eliminate multiple warnings).

>> I still think my suggestion for avoiding the variable consistency is a good
>> one.  I have very rough pseudocode below
>>  
> Done.
>> You could even be smarter in your choice of margins and line-widths to set.
>> 
>> For example, if left_margin > 0 and right_margin <0, you could do
>> 
>> set_paper_layout (left_margin, line_width + right_margin, 0)
>> 
>> or something like that.
>>  
> What do you want to achieve by this?
I want to avoid the multiple warnings you were concerned about.

> I don't really like the idea of
> mixing variables set in \paper {} with computed values from normalize ().
> The only aim of normalize () is to add missing values (although this
> isn't needed in some cases, e.g. right-margin is never used elsewhere)
> and set default values if the settings are considered bad.
> What I would like to add is the possibility to switch off this
> "consistency check", so that you can have negative margins and exceeding
> line-widths if you need to.

As long as the consistency check is enabled, you won't be using  the paper{}
values anyway, so it seems to me that modifying them to create consistency
is no problem.

If you're planning to be able to disable the consistency check only for a
small portion of the output, then I could see that one would not want to
change the paper{} variables as part of the consistency check.

As I said, I don't have strong opinions about this.  You can ignore this
suggestion if it doesn't make sense to you.

HTH,

Carl