Re: [lmi] Generating screenshots automatically

[Vadim Zeitlin]

On Tue, 10 Feb 2009 01:54:12 +0000 Greg Chicares <address@hidden> wrote:

GC> New end users would appreciate annotated screenshots of the various
GC> panels on lmi's tabbed dialogs. Those dialogs are all specified by
GC> 'skin*.xrc'. It's be really nice if we could automatically create
GC> screenshots that use the contents of the WXXRC <help> tags, which we
GC> have already written for every control, as the annotations.

 Unfortunately I don't see how to do this automatically because the
placement of these annotations usually needs to be carefully chosen
manually to make them nicely looking or even just readable.

GC> and the second even displays an annotation as though it were msw
GC> balloon help. Perhaps that's one feasible solution--somehow cause all
GC> balloon help in WXXRC tags to be displayed simultaneously?

 I don't know of any way to do this unfortunately. We could try make to
show each balloon one by one and superimpose the screenshots over each
other though. Should I look into doing this?

GC> Of course there are other ways: e.g., surround the screenshot with
GC> text, formatted in pretty much any manner, connected with lines:
GC> 
GC>   Freeform
GC>   comments
GC>         \
GC>          \ +-----screenshot-----+
GC>           \| [   ] Comments     |
GC>            |                    |
GC>            | ☑ Effective today | \
GC>            +--------------------|  \
GC>                                     \
GC>                                   Synchronize effective date
GC>                                   with current date

 Again, the problem I have with this is how to position the labels
automatically.

GC> Another idea: let the traditional Talmud layout inspire us. Here's
GC> the concept, expressed in English:
GC>   http://keep2.sjfc.edu/staff/mherzbrun/Talmud%20page%20explained%202x3.gif
GC> and in the middle of this page:
GC>   http://ccat.sas.upenn.edu/rs/2/Judaism/talmud.html
GC> we see an intriguing though non-traditional use of color, which can
GC> substitute for lines and arrows: a pretty clever idea, I'd say.

 Yes but I hope LMI controls are less complicated than this...

GC> Here's yet one more idea. Write a program that takes an '.xrc' file
GC> as its parameter and does this:
GC>   for each wxPanel in the file
GC>     render the panel to '.png'
GC>     write out [see below] all <name> attributes and <help> contents
GC> First way: "write out" means render the text into the '.png' file,
GC> below the screenshot of the panel. Second way: "write out" means
GC> embed the '.png' in html as an <img>, and add text as <p>...</p>.
GC> This idea emphasizes ease of implementation (I suppose),

 Yes, this is pretty simple to do.

GC> at the
GC> cost of limiting the scope. Some manual touch-up would be needed,
GC> say, to draw arrows connecting each control to its text.

 A variation: create a DIA file containing the screenshot and the text
boxes with the help text. Then you could open it interactively in DIA, drag
the text boxes around to position them nicely and add arrows as needed. I
suggest DIA because it's the only tool I know but maybe there are other,
better choices too. In particular maybe we could generate SVG -- are there
any nice SVG editors around? I only know of Inkscape (and only because it
uses wx) but I've never really used it.

 Anyhow, as long as we don't have to position the text boxes and arrows
automatically this is certainly doable.

 Regards,
VZ