Re: [Boost-docs] Re: Boost Book Status

[James Fowler]

Joel de Guzman wrote: Hmmm. That's odd. Can you spot the problem? Now that you mention it, I suppose I can... ; )  I was trying to stay within the "stock" QuickBook 1.0 to address Robert's "is it ready" question, and I guess that hit me with tunnel vision.  Which is too bad, because the fix took me less time than trying to figure out how to work around the problem...  In the constructor for actions (somewhere around line 109 in quickbook.cpp), simply change from             , anchor(out) to             , anchor(phrase) and it seems to work much better.  I'd love to claim that some mystical insight enabled me to instantly locate the problem.  Of course, the real story is that as soon as you brought it up, it sounded familiar... actually quite like the problem I encountered working on [sectionlink ...] which, I must confess, took me a bit longer to pin down.  While I haven't dug into the internals yet, it seems that when an action is constructed to target the "out" stream instead of the "phrase" stream, it's output will be written prior to any actions on the "phrase" stream within some higher grouping/category/section/thingy.  It looks like this distinction (out vs. phrase) might not have been originally there, because some of the code looks a little confused (for example, begin_section_action refers to it's stream as "phrase", but is constructed with "out"... and anchor_action calls it "out" when it should have called it, and constructed it, with "phrase").  I didn't see any obvious other likely mismatches in the constructor calls, so this should "fix" things.  At some point, we can document the rationale for "out" vs. "phrase" and clean up the naming inside the actions, although that's not as urgent. So, with the constructor fixed, it appears to be quite happy with what I originally tried: * [#bibliography.1] David Sugar and Daniel Silverstone, "CommonC++ Libraries", [@http://www.gnu.org/software/commoncpp] * [#bibliography.2] Nicola Santi, "Eternity", [@http://www.sourceforge.net/projects/eternity-it/] * [#bibliography.3] Allen Holub, "Roll Your Own Persistence", [_ Microsoft Systems Journal] vol 11, no 6 Jun 1996 * [#bibliography.4] Tasos Kontogiorgos & Michael Kim, "A C++ Template-Based Application Architecture", [_ C++ Report] which looks much more manageable, IMHO, than [#bibliography.1] * David Sugar and Daniel Silverstone, "CommonC++ Libraries", [@http://www.gnu.org/software/commoncpp] [#bibliography.2] * Nicola Santi, "Eternity", [@http://www.sourceforge.net/projects/eternity-it/] [#bibliography.3] * Allen Holub, "Roll Your Own Persistence", [_ Microsoft Systems Journal] vol 11, no 6 Jun 1996 [#bibliography.4] * Tasos Kontogiorgos & Michael Kim, "A C++ Template-Based Application Architecture", [_ C++ Report] ... and invisible underlining (although maybe that's a style sheet issue).  BoostBook problem, AFAICT. I've reported this before. Please correct me if I'm wrong. Could be either BoostBook or the style sheet, I haven't worked with it enough to tell.  However, this excerpt from the generated HTML significant: <li> <a name="bibliography.12"></a> Erich Gamma, Richard Helm, Ralph Johnson, and John Vlissides, <span class="underline">Design Patterns</span>, Addison-Wesley Publishing Company, 1995 </li> <li> <a name="bibliography.14"></a> Vandervoorde and Josuttis, <span class="bold"><b>C++ Templates - A Complete Guide</b></span>, Addison-Wesley, 2003 </li> Notice the first uses <span class="underline">, which doesn't actually seem to make any visible difference,, and the second uses <span class="bold">, which shows up quite nicely.  Of course, a likely explanation is that the second also wraps it's contents in <b>...</b>, and sure enough, manually adding <u>...</u> inside the span for the underline makes the underline show up just fine.  And after all that, perhaps it would have been easier to just try adding span.underline {     text-decoration: underline; } to boostbook.css in the first place instead of writing all this out... but since I wrote all this out already,  you get to read through it all before finding out that this seems to make underline work fine... at least with FireFox : ) - james -- __________________________________________________________ James Fowler, Open Sea Consulting http://www.OpenSeaConsulting.com, Marietta, Georgia, USA Do C++ Right. http://www.OpenCpp.org, opening soon!