Re: my.php.net

--- Gabor Hojtsy <gabor@hojtsy.hu> wrote: > We had gone through many of the things you mention before, and > everything has it's own reason, I can assure you ;) > > > Let's say, for instance, I don't use MySQL very much, but I know there's > a > > function to get a result row, so I open the url: > > > > http://php.net/mysql_get_row > > > > dead end. OK, so I open another URL, http://php.net/mysql > > and I'm closer > to > > where I need to be. > > Well, If you would like to get some info on MySQL functions, you will > probably start from mysql anyway (at least this is what I do ;). > > > Now let's say I'm looking for something a bit more obscure, so I use the > > onpage search input box, and enter 'reload' (just for argument's sake). > > Oops, it defaults to doing a function quickref search; no problem, I > click > > the link to do a whole site search. > > Well this default is what most of the users expect (despite that it is > the same as if they write in the keyword to the URL ;). > > > Well ok, I have some results, but not really what I want, so I'd like to > try > > the word 'refresh' instead. Hmm, neither my last search term is in the > > 'search for: ' box, nor is the type of search I did selected in the 'in > the: > > ' drop-down menu. > > We experimented with this stuff a long time ago. The 'problem' with this > is that the pages are cached, which is actually good for our servers. So > any search term someone enters will be 'remembered' for anybody else > using the same public cache. We had many bugreports about this, and > removed this 'feature'. However I do have plans to adding this again, > but with a different spin. I had a my.php.net RFC here some time ago, in > which I outlined this. > > The solution is to have dynamically loaded small page parts (CSS, JS), > which are not cached. So most of the page can be cached, and user > specific CSS and JS code can enhance the experience (ie. set the search > box to some default, or select the previous search type dropdown element). > > > So I renter my new term, select how I want to search again and make the > > search. Great, ncurses_refresh was what I was after; but wait, the first > > four results are in languages I don't speak, yet I have a language > preference > > set? > > You have selected 'whole site' search instead of 'documentation' search > hey! You are cheating ;) Anyway, doc search is defaulted now to English, > which is not bad most of the time. But it will be enhanced to default to > the preferred one. Why isn't that done so far? > > - It would be very nice to have advanced search features, > like restricting the search to only function documentation > pages in one language, or being able to specify AND/OR > relations between keywords. This all can be done with htdig, > but... > - I didn't have time for it :) > - Most guys say mnogosearch is better, and pl.php.net will > experiment with mnogosearch > - Stig (who works for FastSearch) said he may be able to > set up a central search server at alltheweb for us, which > would mean all the time we spend on the search feature goes > to the trash... > > > I don't mean to be pedantic as php.net is one of the finest resources on > the > > net. However, there have been some annoyances in search and navigation > flow > > that with the advent of setting a cookie I would have hoped to be > eliminated. > > I'm just playing Devil's Advocate in the hopes of streamlineing some of > the > > idiosyncrancies I've found in using the site. > > Well, there are inconsistencies yes. As I have time, I contribute to > fixing them (while discussing problems here too). But the website > project is open for any contributions too ;). Setting a cookie does not > solve anything. We should keep in mind that things should work with > caching, on one hundred mirror sites, for users with all kinds of > browsers (even wget, links and lynx), etc. I'm having some time now and I'd like to contribute to php.net. I'm not very strong with client-side, but know server-side/sql coding and architecture like the back of my hand. I'd also like to take a look at the RFC. Was it posted to systems@ ? If I can get subscribed to systems@ it'd avoid a lot of long Cc lists :) Looking forward to working with you Goba (et al), Hans > > Lastly, in getting a mirror of the site, I noticed that index.php spews a > > couple seemingly needless warnings. I've attached a patch that may be of > > use. > > OK, I'll apply this... > > Goba