- From: olivier Thereaux <ot@w3.org>
- Date: Tue, 2 May 2006 10:07:22 +0900
- To: QA Dev <public-qa-dev@w3.org>
- Cc: Ted Guild <ted@w3.org>, Gerald Oskoboiny <gerald@w3.org>
- Message-Id: <750eb329bab27890bf66fb6ec975d458@w3.org>
On May 2, 2006, at 1:36, Bjoern Hoehrmann wrote: > How about other DTDs, e.g. those for HTML? They also get downloaded, although less so than XHTML ones. attached is a small file giving a good idea (I think) of the frequency at which the DTDs are requested. XHTML1 are the most popular by far, with HTML4 transitional and xhtml-mod DTD second and third. > I would suspect this is due > documents that have an incorrect XHTML FPI and so get recognized as > HTML > documents; since the XHTML DTDs are not in the SGML catalog, OpenSP > will > fetch the XHTML DTD specified in the SI over the network. If that is > the > problem, you would however be able to reproduce it. Hmm, I guess that does not explain everything, but that's a tempting explanation. I'll test it today. If that were the case, would merging sgml.dtd and xml.dtd, and have xml.soc and sgml.soc refer to the same fallback system catalog, help? I can't think of any issue it may cause, but I could be overlooking something. > Likewise, the custom DTD test would have to fetch the DTD; there might > be a problem in your analysis. Right, due to the idiosyncratic nature of logging on W3C mirror, I hadn't noticed that the request for that custom DTD was logged elsewhere. It does get consistently fetched. Thanks for your thoughts! -- olivier
Attachments
- text/plain attachment: dtd_queries.txt
Received on Tuesday, 2 May 2006 01:07:38 UTC