Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

#15 Refresh open new window

BASE
closed-postponed
Tim Rupp
Interface (166)
5
2005-10-02
2005-02-02
Christian
No

When you open for example ”Today's alerts: unique” this
page gets refreshed correctly according to the
$stat_page_refresh_time value in conf. But if you take
one of the alerts and choose to open it in new window
the next time the ”Today's alerts: unique” page gets a
refresh it will only display the alert you opened in the
new window (very annoying).

I tested with firefox and IE and both gave same result.

Discussion

  • Joel Esler
    Joel Esler
    2005-02-02

    • milestone: --> BASE
     
  • Joel Esler
    Joel Esler
    2005-02-08

    • assigned_to: nobody --> caphrim007
     
  • Joel Esler
    Joel Esler
    2005-02-08

    Logged In: YES
    user_id=853584

    Tim, this has something to do with that session thing you working on.. Is
    this the same thing?

     
  • Tim Rupp
    Tim Rupp
    2005-02-08

    Logged In: YES
    user_id=628253

    Yes Joel I think the two are related. When moving between
    pages, lots of session variables are set (see $debug_mode =
    2). The problem is that lots of these session variables have
    no relevance to data between pages. So when you change a
    session variable on one page, it will carry over to every
    other page until it is overwritten by another value. And
    other pages may use variables that are named the same as
    that in the session.

    I believe we should be using GET and POST and NOT session
    vars to control what pages get sent what (*slaps Roman's
    wrist* bad Roman, go to your room). By doing this, you
    isolate what is sent between pages and you ensure that a
    value in one variable, will be destroyed between page surfs.
    This should eliminate these problems.

    Now the only hard part is actually going through and finding
    all the places where the session data is used and read, and
    replacing it with calls to ImportHTTPVars or whatever. The
    session stuff is just so dug into the app that it's going to
    take a good chunk of time on my end to weed it all out.

    I'm planning on setting up a separate copy of base that I
    can brutally hack into and make changes on without tainting
    cvs. Once I'm able to get it online, I'll pass around a link
    and ask everyone to thoroughly test it before I make any
    commits.

    One thing that will need to be addressed though once I can
    change this around is input validation and making sure we
    never create a url longer than the maximum length set by
    most browsers (2083 characters??? I could be way off on that
    but I do know there is a limit). I've seen base make some
    pretty long urls but I dont see this as a big big issue,
    input validation would be bigger.

    Anyways, lemme set up a separate instance of base and start
    working on this session stuff. It'll take a while I predict
    but I think it will make at least the 1.x tree a lot more
    navigation bug free.

    Tim

     
  • Joel Esler
    Joel Esler
    2005-02-28

    • status: open --> pending
     
  • Joel Esler
    Joel Esler
    2005-02-28

    • status: pending --> pending-fixed
     
  • Christian
    Christian
    2005-02-28

    Logged In: YES
    user_id=1155549

    changed status since its not fixed. Time is aware of the
    problem and will have a look when he got time.

    /Christian

     
  • Christian
    Christian
    2005-02-28

    • status: pending-fixed --> open
     
  • Joel Esler
    Joel Esler
    2005-05-26

    Logged In: YES
    user_id=853584

    Tim,

    Did you ever get a chance to do what you were talking about
    here?

    J

     
  • Kevin Johnson
    Kevin Johnson
    2005-10-02

    • status: open --> closed-postponed
     
  • Kevin Johnson
    Kevin Johnson
    2005-10-02

    Logged In: YES
    user_id=836228

    This will be fixed in 2.x since it requires the entire
    session handling to be rewritten