Can anyone get the EMBT docwiki search to provide a useful hit?
It seems broken to me.
Marco Cantù – I am starting to suspect that someone has stopped Google from indexing the whole site – which is a really bad idea when you use a Google Custom search to search it.
Lars Fosdal http://docwiki.embarcadero.com/robots.txt
There are some places not allowed.
Lars Fosdal Working for me. Can you explain what you were searching and what hits you expected?
I navigate to the XE8 page, then search for “HintPause”
TApplication.HintPause – “Did you mean TApplication.HInstance” ?
Seems the search is context sensitive. You need to go the Libraries Reference page: http://docwiki.embarcadero.com/Libraries/XE8/en/Main_Page and do the search there
I generally do a standard google search with “delphi docwiki” prefix, and can find what I’m looking for. The robots file is used to push google to index (or give priority) to recent content, compared to old one. Given the same pages (a few tens of thousands) exist for each of the versions of the product, better if developer hit the latest…
http://docwiki.embarcadero.com/RADStudio/XE8/en/Main_Page fails the search. I guess it would have been better if it suggested the Library page from that fail?
delphi docwiki works well, though.
Thanks!
Marco Cantù When googling for Delphi I usually get Delphi 2009 docs. You may want to remove those from being indexed.
For instance Googling for Delphi TObject.GetType returns http://docs.embarcadero.com/products/rad_studio/delphiAndcpp2009/HelpUpdate2/EN/html/delphivclwin32/System__TObject__ClassType.html
Dalija Prasnikar That is why Marco suggested adding “docwiki” to the search. Personally I added inurl:docwiki to my custom search engine you can define in Chrome.
Go to chrome://settings/searchEngines and then add your DocWiki search (i took dw as keyword) with this address: http://www.google.com/search?q=%s+inurl:docwiki
Then I can type dw TApplication.HintPause in my address bar hit enter and boom have the correct link to the XE8 doc.
Stefan Glienke Yeah, adding docwiki works, but you have to know that first in order to use it.
Marco Cantù The robots.txt file is not supposed to be used to prioritise content. In fact having many pages with near-duplicate content, which you do, is known to cause problems with Google’s algorithm and de-prioritises the entire domain when giving back search results – which is bad. It’s because Google sees many near-duplicates and concludes it’s a spam site. Ie you may inadvertently be causing Google to not value the docwiki as good content at all. That’s probably why you have to prefix your search with “delphi docwiki” whereas with a well-configured doc site, a search should show the docwiki in the first page of results, and show XE8’s entry highest out of all Delphi versions.
The correct way to do it is to add tags to older content – ie the D2009 to XE7 pages – indicating the “canonical document”, that is, the latest version of the same content. That would point to the XE8 entry for the same thing. You can read more here: https://support.google.com/webmasters/answer/139066?hl=en and here: https://support.google.com/webmasters/answer/139066?hl=en#2
Ie, let all content be indexed, but for old content, add a tag that tells Google that there is a newer version of the “same” page. It stops it concluding the site is full of spam, returns the site higher in search listings, and will always prioritise the newest (ie XE8, currently) version of any page over, say, the D2009 one, when returning search results.
I sent an email to the doc people about this quite a while ago, in the XE6 timeframe, but did not get a reply. (If you want to ask them to look it up, it was sent to documentation@embarcadero.com on 24/05/2014, subject “Canonical versions of old pages on the Delphi docwiki”.)
I have no idea if the docwiki infrastructure we are using has that support, if not it will be quite difficult to add it. We don’t suffer from Google penalties on the dup content, not that I know of. Adding “Delphi docwiki” is generally to filter out non Delphi content and some web sites with higher hits (including our own other sites). In many cases it is not needed. But if you search for a method that exists in .NET, you can guess where you are re-directed.
If you can resend me the email, I can forward to the right person (not the same), and have a chat about it. But if the content is the same of above, I can forward this thread.
The content is basically the same, except in the email I also asked for a human-readable link to be added, something like “See the latest version of this page”. So if you’re reading the XE2 docs you could click and visit the XE8 version of the same page instead.
It’s a pity if it’s difficult to add that. Internally, is there any linkage knowing that one page is the “same” as another, only for a different version? If not can you guess based on path? If so it should be possible to automatically create those tags in the generated page output – and it would be a very good thing to do so.
It is a little bit weird (though understandable) that you have a direct link to the same page for lower Delphi versions, but when you are at a lower level you have no link upwards. For example in XE8 you have all the Delphi versions from XE7 downto D2010 on the left, but when you click on say XE7 you have just XE6 – D2010, but no clue that there might be newer content for XE8.
The docwiki results are terrible. I use the ‘delphi docwiki …’ trick too. Embed the Google search results for free above the docwiki results. https://support.google.com/adsense/answer/9879?ctx=as2&hl=en&rd=2&ref_topic=1705820 You can set it up to do site:docwiki.embarcadero.com automatically. Or whitelabeled (but not free): https://cse.google.com/cse/
I always use Google to search, never built in searches. Too bad that only works for public sites available without having to login.
maybe emb could add a drop down option to search with google (add custom google search to your site) :-
https://support.google.com/customsearch/answer/2630969?hl=en
Brett, they are already using a custom Google search.