yaws

Hi!

In case that anybody is interested...

On Wed, Feb 11, 2004 at 01:50:43AM +0100, Carsten Schultz wrote:
> On Mon, Feb 09, 2004 at 04:10:15PM +0100, Carsten Schultz wrote:
> > I have checked in code to send deflated content.  To activate it, you
> > have to set `deflate =3D true' in a server config.
>=20
> Please use this with care.  There seem to be a lot of browsers out
> there which implement this incorrectly.

Since there were more browser problems than I had anticipated, I have
switched to `gzip' encoding.  The option to turn it on is still called
`deflate' however.  So if you want to give it a try, I am quite
confident that it won't break your site.  If you want to see it in
action, you can have a look at http://www.fu-mathe-team.de, I have enabled
compression for it.

Compression is disabled for browsers which I know to be broken (see
yaws:has_buggy_gzip/2).  If you want to test your browser
independently of this, you can have a look at
http://carsten.codimi.de/gzip.yaws/ for a test of gzip and deflate
which does not use the new Yaws code and does not have compression
disabled for buggy browsers.  I am interested in your findings.

Same things apply as before: You need to have the zlib module (which
is in R9C or jungerl, for example).  Only static files are compressed.
Since large files are already compressed dynamically, it will not be
too hard to compress dynamic files; I plan to do this, but time does
not allow it for some weeks.  If you think that the current code is
usable, I will document it, however.

Greetings,

Carsten

--=20
Carsten Schultz (2:38, 33:47), FB Mathematik, FU Berlin
http://carsten.codimi.de/
PGP/GPG key on the pgp.net key servers,=20
fingerprint on my home page.

>>>>> "klacke" == klacke  <klacke@...> writes:
>
>
> Ok, both patches applied. The first one modified though. 

Great!

> - I didn't like the lists:member call, replaced with
>   faster func clause.
> - I added the br tag (Was that correct ??)

Ah, yes, that was good.

> - I applied the same procedure for the ehtml_expander (the
>   ehtml template expander code)


> Question: where did you get the list of tags from ??

I did a quick scan through the HTML 4.01-spec, mainly
looking for tags for which I thought that a \n before the
tag would actually change the appearance. For example, for
sup it would, but for h1 it would not since they appear on a
separate line anyway. Thus no systematic approach, I'm afraid.


/Tomas

Hi,

	I set out to convert the wiki in yaws into ehtml[1],
as well as to add a few features from another wiki (from one
that is based on Joe's wiki/pico).

However, I noted a couple of things about ehtml and yaws
that I thought I should share:

The first one is that yaws_api.erl unconditionally prepends
tags with \n (yaws_api.erl, around line 1223). While this
makes the rendered html code nice and easy to read, it also
introduces a whitespace, which can be significant in some
contexts[2]. I've attached a patch which attempts to fix
this in a reasonable way, I think.

The second is that when setting up more than one virtual
server on the same IP and port number, I lacked a way to say
that a certain server config is to be selected if no server
config matches the Host:-field in the HTTP-request (someone
may have typed in e.g. http://123.123.123.123/).  So I added
an is_default option to the server config. I've attached a
patch for this (the patch also includes documentation
updates).


I hope to finish the wiki updates soon as well, but I'd like
to discuss them first (in another posting).


/Tomas


____________
[1] mainly because I wanted a parser for the wiki
format (wiki_format_txt), that could be reused in
other environments and maybe with other backends
as well (LaTeX comes to mind).

[2] Example: {ehtml, ["E=3Dmc",{sup,[],"2"}]}
will prodcuce "E=3Dmc\n<sup>2</sup>", which the browser
will rendered as "E=3Dmc =B2". It looks better without
the space in this case.