From: <wi...@us...> - 2007-06-04 13:10:44
|
Author: wiemann Date: 2007-06-04 15:10:33 +0200 (Mon, 04 Jun 2007) New Revision: 5207 Modified: trunk/sandbox/package-doc/multiple-input-files.txt Log: some fixes/additions Modified: trunk/sandbox/package-doc/multiple-input-files.txt =================================================================== --- trunk/sandbox/package-doc/multiple-input-files.txt 2007-06-04 13:02:52 UTC (rev 5206) +++ trunk/sandbox/package-doc/multiple-input-files.txt 2007-06-04 13:10:33 UTC (rev 5207) @@ -519,21 +519,21 @@ As described in section Caching_, when processing files stand-alone and resolving their external references, it may be necessary to -process or re-process referenced files. Since this is during +process (or re-process) referenced files. Since this is during transform-time, the parser instance is no longer available; it is -therefore necessary to create a new instantiate a new parser -instance. +therefore necessary to create a new instance. This instance has to be parameterized somehow, for instance to expect -PEP syntax or a normal reStructuredText file. There are three +PEP syntax or a normal reStructuredText file. There are four possibilities to do this: File name guessing (recognize pep-\*.txt), content guessing (recognize an RFC 2822 list at the top with a PEP: -line), or configuration via docutils.conf. File name guessing should -do for now. +line), configuration via docutils.conf, or always assuming standard +reST (no support for PEP's). Either file name guessing or only +support for standard reST should do; they are easiest to implement. All requests for doctree and name-to-ID mappings should go through the caching system. In case of a miss, the caching system instantiates a -parser and re-parses the requested file. Perhaps the caching system +parser and (re-)parses the requested file. Perhaps the caching system should also be responsible for file name guessing (alternatives: the standalone reader, the reStructuredText parser), and for deciding which parser class to instantiate. @@ -541,7 +541,7 @@ Also, all calls from the publisher should be routed through the caching system. In the case of independent files which are not part of a larger documentation tree, the system simply does not find a -project root and assumes a cache miss. +project root and always assumes a cache miss. This also means that no separate front-end is needed anymore for PEP files, since they are recognized by their file names. |