Re: cvs log: converted m4 sources to canonical format

[Guido Draheim]

Peter Simons wrote:
> Guido Draheim writes:
> 
>  > I guess that the tag contents are not allowed to be
>  > free-format #pcdata ?
> 
> No, @category and @license both want registered keywords.
> Those keywords can be arbitrary, though.
> 
> 
>  > Why not using the names as visible, containing spaces
>  > like "C Support"?
> 
> Allowing the submitter to specify an arbitrary category name
> is, in my experience, an invitation for spelling errors. If
> you write "AutoMack" instead of "Automake", then the
> software will notice it immediately. If whatever you specify
> at @category becomes the literal name of the category, then
> a human _might_ notice we have "C Suppart" now when he looks
> at the generated HTML page.
> 
> 
>  > Do people have to do an inverse-lookup to know what
>  > category/license hint text they have to add into the
>  > macro text before submission?
> 
> The submitters can write into that field whatever they want,
> or they can just leave it blank. What counts is what the
> maintainer who commits the macro into CVS writes into that
> field. Getting that stuff right is our job, so to speak, and
> the software helps with that by drawing attention to any
> suspicious content (because it knows which categories
> exist).

This is bogus - the repository file should be no different
from the download file which should be no different from
submission file - allowing anyone to modify a downloaded
file and e-mail it to the author forwarding it back to the
ac archive.

Sure you may want to check the category item to list - but
why do those items in the list need to be any different
from the list visible on the website? Is there _any_ need
for that? I was guessing the section was errornously
modelled as an #id-attribute but if it can be of any
format then it can be the one on the webpages - containing
spaces and ultimately looking like
@category C Support

And please, no need to check spelling errors in the
intermediate xml file as long as one can check spelling
errors during parsing. If all you have a hammer...

> 
> I'll make sure, though, that the list of existing categories
> is always right next to the submission instructions. ;-)
> 
> Anyway, I am open to suggestions for categories that do we
> still need. "Python" has already come up, and I think the
> entries
> 
>  > System Headers
>  > System Utilities
>  > Latex
> 
> from your list are also good ideas.
> 

The earlier two were agreed upon a year ago but it seems
they were not being used so far. The latex one is for
single submitter creating a long series of macros that
were also orthogonal to the usual ac macros detecting
a (c/c++) compiler environment.

-- have fun, guido