[Edbrowse-dev] workarounds

[Karl Dahlke <eklhad@comcast.net>]

The last few commits work around some tidy issues that we're still looking into.
Like other workarounds in the past, I may be able to delete them at some point.

I use to preprocess the html to work around a <script> parsing bug,
the bug now fixed, and I took out that code, but left the hook in place
so we can put in other preprocessing code as needed.
Today the preprocessor turns < into &lt; and > into &gt;
if those characters appear in a textarea,
as tidyh will treat them as tags and they're not.
Go to http://www.prchecker.info and put in a url and check it,
next page has a textarea full of html tags.

Other is not a fix but just something -
sometimes an anchor is closed early, and has no text inside.
This looks like {}, something you can click on but no text inside.
Oddly enough I had code in format.c specifically to remove these,
I really don't know why,
so #if around it and now {} appears,
sometimes, and if you're not sure,
odds are the text following is the text that should be inside,
and that tells you what the link is.
It might be possible to rearrange the node tree so that text is inside again,
but that's a lot more work and maybe there's a tidy fix for this one
so I'll hold off on that.

Sometimes workaround has to happen before, sometimes after.

On the ls world, lsy gives you the path if the current file is a symlink.
I had length, time, major minor, inode, everything,
except symlink, and that's what I wanted the other day when scanning through /lib
and I wanted to make sure the .so was linked to the right thing etc.
This probably won't be used vary often, and has no meaning in windows.

Geoff still trying to port mozjs to windows, and apparently it's not easy.
It looks like duktape could be an easier port so something else to think about.

Karl Dahlke