How can a so called "stable" application base itself on a generated page ? Why don't you checkout the XML sources of the docs and analyze the XML
directly ?
The XML sources contains what you are asking for (the mysql_XXXX() functions for example are under the reference/mysql/functions/ directory)
Mehdi
Yes, 'so called "stable"' would be the word, since I'll probably need to supply patches when new manuals come out. I'd kinda thought, why add the howemany megs to this app and burn up more bandwidth? However, you do have a good point. As it stands, it now just scrapes the downloadable multi-page html manual and seems to do a good enough job at it, it seems, I'll just have to offer the version of the manual I now have if there are any formatting changes, or else offer a patch.
The question is about these directory names (which are sort of like the function family or category): nowhere do docs imply that these are any more than just directory names that can change from one copy of the manual to the next. However, they're pretty good as they stand as reference names most of them are also used in many function names, so I suppose they're considered somewhat stable. Question is basically: are they stable, yes or no? could they be made or declared stable if this hasn't really been though out before? E.g., functions in the chapter title 'Ingres II' are in the directory 'ingres' - this is nice, doesn't have any confusing spaces in it, suppose Mr. Ingres marries Ms. Cohen and changes his db name to Ingres-Cohen PowerDB, the "normal language" name of the category would change while the reference name ('ingres') would stay the same.
Cheers, James
J.M. Coder wrote:
I mean the general category or type of the function - i.e., mysql_num_rows() is in the Mysql family of functions, with the linkname 'mysql' (from ref.mysql.php). I like this, no problem - what I wonder though, I'm writing something which produces some info based on the manual, and one of the things it says about a function is which type of function it is (string, array, mysql, etc). It' be nice if the type classifier could be formalized, this would help apps that produce info from the docs be more 'stable'.
-James
Gabor Hojtsy wrote:
Thx to Sean Coates for his suggestion re. getting synopsized info from the manual.
I have a small suggestion after looking at Sean's excellent site phpdoc.info and looking at the schema DocSkel, http://wiki.phpdoc.info/DocSkel/FunctionskeletonDotXml?v=103j
At the moment, functions are divided into separate chapters, or sections, of the function reference, with real-language sounding names in the anchortext and [a-z0-9]+ names in the links, which differ from the real-language names. Real language names usually don't make efficient classifiers for automated tasks, w/spaces and alternate spellings and all, but I as yet don't see any officially sanctioned classification names. I'm not familliar with the workings of the docs generation process, but could I suggest, why not just officialize the current [a-z0-9]+ function family names as they're found in the manual links? This would give people who are writing stuff for manipulating/presenting the docs a bit more of a permanent way to refer to them. Otherwise these names are just fairly arbitrary words in the links, which can change without much thought. If there are official classification names floating around somewhere, would appreciate a headsup.
Function names, like mysql_num_rows becomes mysql-num-rows in the URL, this is formalized. What is the problem with this, or what else would you like to have? Would you please provide examples?
Goba
