
Marcello Perathoner wrote: [restatement snipped]
It is not! (People should really read the specs!) If it were #PCDATA (*Parsed* Character DATA) the validator would find the error because it would parse it. Its because its CDATA that the validator doesn't parse it and in consequence cannot find the "error". In CDATA < has no special meaning at all.
True. In my original analysis I was looking at the DTD declaration for file number 15701 which is XHTML Strict as opposed to file number 15698 which is indeed HTML 4.01 Transitional. But in this case the distinction between #PCDATA and CDATA is inconsequential, because for this bug to manifest itself we are dealing with a User Agent which either does not support the <style> element, or has implemented that support incorrectly (as in the case of Opera 7.11). Thus, my analysis, and your restatement, are still correct. Hopefully, this little quibble will not distract people from the larger issue, which is that the file at PG is flawed, and needs to be corrected. David A. Desrosiers wrote:
The balance has to be consired: Is it worth it to incur another socket connection and round trip from the client (more server hits to retrieve resources) or is it better to have a larger byte-per-hit (but less socket connections) per-client?
This objection, and Marcello's response, are both based on a faulty assumption: that the primary use of HTML files will be online, served up by some sort of HTTP server. I would bet that the vast majority of all HTML files offered by Project Gutenberg are downloaded to a local computer, and then read while offline. I know that _I_ have never read a PG e-text directly from the web server. Fetching multiple files from the local file system is simply a non-issue. And if I want to substitute my own styles for those of the original poster, downloading to a local file system is required. If you're going to insist that user selection of styles is not possible, the question of whether they are internal styles or external styles is irrelevant, but if you're going to permit user flexibility in selecting styles, the issue of multiple GETs is irrelvant. This is one of the reasons that it is important that all HTML files be offered in ZIP format: because it is likely that a user will be reading offline, s/he should be able to fetch all the files associated with a work (styles and images) in a single download.