06 January 2005

Search, PubSub, 6-Jan-05 UTC 21:57:14

Curious quirks with PubSub:

  • updated searches include deleted parameters

  • Auto-adjusting searches; not recognizing http://,

    The long voyage

    What’s interesting about PubSub: You can have a saved search that returns information. Great, no worries.

    But then if you want to change that particular PubSub subscription by modifying the "search terms", something unusual happens.

    I have a test search with known, specific terms in PubSub. It sends me back things that I know are there. This tells me whether PubSub is working correctly. It does.

    Normally, nothing unusual happens.

    In fact, its become so predictable, that I’ve contemplated removing this “check subscription" because I keep getting good news.

    Until today.

    The original search was producing a number of items that were "outside the range of what I wanted".

    I decided to adjust something. Yes, the old "Oh, no…not that" What I did wars change the search routine-terms in the particular subscription to adjust the range of the search.

    I eliminated a number of terms for the list in the particular PubSub subscription.

    When I normally change in-put criteria to a search, I normally get back a different search output. For example in Google, if we adjust terms, we get a change in the output.

    This is where there interesting part happens.

    For some unusual reason, PubSub continued to return in the subscription-report information that was related to the terms “no longer contained in the subscription.’

    In other words, even though I had updated the subscription terms, saved them, and PubSub returned a confirming message that the changes ‘had taken,’ the results were still linked with the ‘pre-change subscription.’

    I suppose this has something to do with ‘the time needed to update.’ But what is curious is that when I’ve submitted information, both PubSub and Technorati have been fairly quick in returning the updates.

    Not this time. It’s as if the subscription-terms are stuck in a time warp. My terms are updated, but the output is linked with the non-saved terms.

    Strange. I’m not the only one apparently in a time warp.

    Which part of the URL is needed

    The other unusual thing I found with PubSub was ‘what was an acceptable search term.’ Suppose I want to find links for a particular website. You know, see if someone has mentioned in an article [text-wise] a URL-URI.

    In the ‘search world’ of end-users, the ‘normal thing’ to do is to enter a code like http://www..... [and then the rest of the URL-URI].

    If you in put the complete URL code of http://.... [blah, blah]…guess what: PubSub comes back with a message that says, “

    Unrecognized attribute name: http [In red]
    For some reason PubSub does something different. I’m not sure whether it strips out the http as a “required element to find a site’… or what.

    What has me wondering is that if I want to find something like “Hamburger’, PubSub doesn't’t return ‘burger’…so If I am looking for http://www.sitename.com... I’m not sure whether [because the http:// can’t be entered], whether PubSub will not recognize requests for sitename.com…

    Another mystery.

    At this point I'm wondering:

  • If I want to find http://www.sitename.com … will PubSub know the difference between sitename.com and http://www.sitename.com …
  • If PubSub can tell the difference between http://www.sitename.com and sitename.com … does this mean that PubSub knows to strip out http:// without me having to include it?
  • If PubSub “does something’ with http://www … and strips it out, or returns links even though it is not there, are there other things that PubSub strips out that I might want to know about?
  • For example, if I’m searching for something that PubSub automatically adjusts, I’d like to know that … then I wont be waiting for something that PubSub will never give me. I may want only those sties that have the http://www.sitename.com listed as a complete URL, and not want those sites with just sitename.com.

    Also, if PubSub is stripping out things before or after / / those forward slashes, then I’m wondering if I have a strange URL without traditional URI-URL codes, that if PubSub strips something out [that should be left in], I may never find my desired site.

    For example, I’ve seen some strange URL-URI’s in the adobe.files; and also some of the Developer-file-codes have some nifty // forward slashes. Plus those wikis, they sometimes have strange names with funky URL-URIs in them.

    I’m kind of wondering whether there is a pool of URLs and URIs that PubSub may automatically adjust, and this is narrowing my field of focus and search. Something may get missed.

    I sense I shall lose much sleep over this. ;-)

    Curious quirks with PubSub:

  • updated searches include deleted parameters

  • Auto-adjusting searches; not recognizing http://,

    The long voyage

    What’s interesting about PubSub: You can have a saved search that returns information. Great, no worries.

    But then if you want to change that particular PubSub subscription by modifying the "search terms", something unusual happens.

    I have a test search with known, specific terms in PubSub. It sends me back things that I know are there. This tells me whether PubSub is working correctly. It does.

    Normally, nothing unusual happens.

    In fact, its become so predictable, that I’ve contemplated removing this “check subscription" because I keep getting good news.

    Until today.

    The original search was producing a number of items that were "outside the range of what I wanted".

    I decided to adjust something. Yes, the old "Oh, no…not that" What I did wars change the search routine-terms in the particular subscription to adjust the range of the search.

    I eliminated a number of terms for the list in the particular PubSub subscription.

    When I normally change in-put criteria to a search, I normally get back a different search output. For example in Google, if we adjust terms, we get a change in the output.

    This is where there interesting part happens.

    For some unusual reason, PubSub continued to return in the subscription-report information that was related to the terms “no longer contained in the subscription.’

    In other words, even though I had updated the subscription terms, saved them, and PubSub returned a confirming message that the changes ‘had taken,’ the results were still linked with the ‘pre-change subscription.’

    I suppose this has something to do with ‘the time needed to update.’ But what is curious is that when I’ve submitted information, both PubSub and Technorati have been fairly quick in returning the updates.

    Not this time. It’s as if the subscription-terms are stuck in a time warp. My terms are updated, but the output is linked with the non-saved terms.

    Strange. I’m not the only one apparently in a time warp.

    Which part of the URL is needed

    The other unusual thing I found with PubSub was ‘what was an acceptable search term.’ Suppose I want to find links for a particular website. You know, see if someone has mentioned in an article [text-wise] a URL-URI.

    In the ‘search world’ of end-users, the ‘normal thing’ to do is to enter a code like http://www..... [and then the rest of the URL-URI].

    If you in put the complete URL code of http://.... [blah, blah]…guess what: PubSub comes back with a message that says, “

    Unrecognized attribute name: http [In red]
    For some reason PubSub does something different. I’m not sure whether it strips out the http as a “required element to find a site’… or what.

    What has me wondering is that if I want to find something like “Hamburger’, PubSub doesn't’t return ‘burger’…so If I am looking for http://www.sitename.com... I’m not sure whether [because the http:// can’t be entered], whether PubSub will not recognize requests for sitename.com…

    Another mystery.

    At this point I'm wondering:

  • If I want to find http://www.sitename.com … will PubSub know the difference between sitename.com and http://www.sitename.com …
  • If PubSub can tell the difference between http://www.sitename.com and sitename.com … does this mean that PubSub knows to strip out http:// without me having to include it?
  • If PubSub “does something’ with http://www … and strips it out, or returns links even though it is not there, are there other things that PubSub strips out that I might want to know about?
  • For example, if I’m searching for something that PubSub automatically adjusts, I’d like to know that … then I wont be waiting for something that PubSub will never give me. I may want only those sties that have the http://www.sitename.com listed as a complete URL, and not want those sites with just sitename.com.

    Also, if PubSub is stripping out things before or after / / those forward slashes, then I’m wondering if I have a strange URL without traditional URI-URL codes, that if PubSub strips something out [that should be left in], I may never find my desired site.

    For example, I’ve seen some strange URL-URI’s in the adobe.files; and also some of the Developer-file-codes have some nifty // forward slashes. Plus those wikis, they sometimes have strange names with funky URL-URIs in them.

    I’m kind of wondering whether there is a pool of URLs and URIs that PubSub may automatically adjust, and this is narrowing my field of focus and search. Something may get missed.

    I sense I shall lose much sleep over this. ;-)

    " />