Am 07.04.2016 um 07:54 schrieb Urs Liska:
Am 07.04.2016 um 07:50 schrieb Noeck:
Hi Urs,
a very uneducated guess: Could it be that the html pages are only
generated if the input has changed and as you change the generating code
and not the doc-input, the generation is skipped? It is still strange
that it works if you empty the file.
Sorry, I had intended to mention that. In order to check exactly this I
introduced an arbitrary change in a snippet (added a comment). Then I
can see the modified snippet, so obviously it *does* get processed.
My suspicion is that there is some completely different file responsible
for generating the HTML.
Urs
Just after sending this message I had another idea and checked for .pyc
files.
It seems that python/book_html.py does *not* have a .pyc file while
python/out/book_html.py *does*.
However, when deleting the .pyc file and rerunning make doc (with the
emptied book_html.py) the .pyc was silently (and empty) recreated.
However, when *deleting* the .py and .pyc files an error was triggered,
when lilypond-book tries to import book_html.
Looking further into it I realized:
- book_base keeps an array with "formatters"
- upon importing any specific book_X it registers an instance of the
formatter class in book_base
- So if book_html is empty no HTML formatter is registered with
lilypond-book
This is why there is no error with the empty file.
However, this isn't an answer to my question yet. If the HTML formatter
is *not* used to produce the HTML included in the doc pages, what else?
Any further ideas?
Urs