Validation performance in Batch Wizard with multi-core CPU

[Albert Wiersch]

I'm considering, for a future major update (v10 or above), to make CSE HTML Validator take advantage of multiple cores when using the Batch Wizard in the professional edition. This should significantly speed up validations when using the Batch Wizard. However, I'm not sure it's worth the effort. I am thinking the time may better be spent improving the validation and supporting new standards.

Comments? Do you do a lot of Batch Wizard processing on a multi-core system that you wish was much faster?

[MikeGale]

This would certainly be useful for me.

I'd like that.

[roedygr]

My priorities would be in order:

1. invent a fast way of loading all file with problems detected in last batch run ready for edit

2. avoid rescanning html that has not changed since the last batch wizard run. Cache the results. I think this would give me an order of magnitude speed improvement.

3. dual core threads

4. proof the style sheets and style markup.

5. give me ways of automatically fixing simple errors, e.g missing </li>

[roedygr]

and 6. would be getting ready for HTML 5

[MikeGale]

For getting a really crude estimate of the HTML5 timeline look at this

http://a.deveria.com/caniuse/#agents=Al ... cr,wd,ietf

NOTE: It's all "Working Draft" for HTML 5 and the people working on the standard seem to be saying that it will only be really ready in something like 2022. Having said that even HTML 4 does not yet fall into their definition of "really ready", so expect something earlier than that.

I think that there are things missing from the list.

The process by which standards are set, in the past seems to often reflected a lower IQ than any individual on the committee, some standards may have been set vindictively (deliberately ambushing the early implementors!), and there has been a lack of creativity and ambition. Hopefully this effort (which has broken free of how it was done before) will be better.

Three things (from that page) look worth considering:

Editable
X-Document messaging
Cookies++ (The beyond cookies technology for storing name value pairs.)