Re: [MirageOS-devel] using cow

[Anil Madhavapeddy <anil@xxxxxxxxxx>]

On 23 Dec 2013, at 11:59, Richard Mortier <Richard.Mortier@xxxxxxxxxxxxxxxx> 
wrote:

> i'm having a few issues using cow to recreate mirage-decks and wonder whether 
> i'm missing something really obvious. code at 
> <http://github.com/mor1/mirage-decks>.
> 
> each slideshow on decks is served up using reveal.js which itself ends up 
> using an in-browser markdown renderer to render slides from markdown 
> fragments between <section></section> tags.
> 
> at the moment, each deck is in its own directory containing an (incomplete) 
> index.html file which contains the slides as <section /> fragments plus all 
> the images referred to. when a deck is requested i then read the relevant 
> index.html file in, and insert it into a template which is an <:html< ... >> 
> fragment.
> 
> it's this insertion that's causing me problems. if i insert into the template 
> as the anti-qutoation $str:deck$ then all the html in deck is escaped (< 
> becomes &lt; etc). obviously this doesn't render correctly. if i read in the 
> index.html and parse it, then i can insert as $deck$ and that works but it's 
> meant i've needed to do a fair amount of tidying up of the index.html files 
> for all the slide decks to get past cow's parser.
> 
> so-- is there a way to say "just insert this string verbatim" in a quotation 
> (or anti-quotation or whatever the things between $$ are called in cow)?

You can't insert the string verbatim, as the type of the $$ must be a 
Cow.Xml.t.  The whole point of $str:$ is to take care of escaping.

However, you should be able to just parse the template fragment into Cow.Xml.t 
and insert it as such.  I'm guessing it doesn't parse because of...

> fwiw i've had similar issues with literal js between <script> </script> tags 
> which i've overcome by pulling it out into a file and then serving that 
> appropriately. i could probably do something similar here by just having the 
> template as html fragments that are manually prepended/appended, but that 
> feels like a kludge -- given that type-checking is good, being able to 
> typecheck the template as valid html, even if i can't type-check the entire 
> page, seems a good thing... :)

And here you hit the biggest drawback with Cow's parser: the $ anti-quotations 
mean that there's no way to escape them (that I know of).

So if you have:

<:html<
  <script>
    var foo = $bar;
  </script>
>>

You need insert the dollar without typing it in.  I use this:

let d = [`Data "$"]
<:html<
  <script>
    var foo = $d$bar;
  </script>
>>
        
Alternatively, you could make it valid XHTML by wrapping it in a CDATA.

<:html<
  //<![CDATA[
  <script>
    var foo = $bar;
  </script>
  //]]>
>>

The latter is probably the best solution, since we serve the MIME type
as XHTML anyway.

-anil
_______________________________________________
MirageOS-devel mailing list
MirageOS-devel@xxxxxxxxxxxxxxxxxxxx
http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel