Re: Screenshots

[Ben Pfaff]

On Mon, Nov 30, 2020 at 12:00 AM John Darrington
<john@darrington.wattle.id.au> wrote:
>
> On Sun, Nov 29, 2020 at 09:44:45PM -0800, Ben Pfaff wrote:
>      On Sat, Nov 28, 2020 at 1:10 PM John Darrington
>      <john@darrington.wattle.id.au> wrote:
>      >
>      > I've pushed a change which adds some screenshots to the user manual,
>      > along with a framework to autogenerate them.  The sad reality is
>      > however, that while it works automatically, I think it is not ever
>      > going to work in a portable fashion.  So I've checked in the generated
>      > images.
>      >
>      > To regenerate them on a debian system the procedure is:
>      >
>      > apt-get install xdotool xvfb imagemagick x11-apps
>      > make doc-make
>      > make -f doc-make
>
>      That's pretty great! Is the rebuild procedure documented somewhere?
>      Otherwise it will be hard to find this email later.
>
> There are some comments in doc/doc-make.in, but otherwise there is no
> documentation.

Perhaps at least the above should be mentioned somewhere. Maybe
INSTALL would be an appropriate place.

>      For myself, I'm working to get better output into the manuals. I'm most
>      of the way toward getting good .png output from the cairo driver. I think
>      that I've done the rest of the work already.
>
> Using PNGs for pspp output will be okay, for the printed manuals,  but not so
> good for somebody using the standalone info reader, or for somebody browsing
> the HTML using a text only browser.

Standalone Info users will still see the text version, since Info has special
support for that case. I don't know whether the HTML output by makeinfo has
such a fallback, but if not, anyone who uses a text-only browser these days is
probably savvy enough to look at the output .txt themselves, or to switch to
the Info version.

The HTML version of our output doesn't look very good. Another item on
my to-do list is to improve its look, and then make the HTML version of the
manual use that instead. Then the text-only browser users would see
something OK, or at least as well as a text-only browser renders tables
and CSS (which is, admittedly, usually not very good).