Yes - the suspend/resume idiom is correct.
I've just put together something to read the data asynchronously 
as a per-domain option.  I'll stick another 3.2 "snapshot" up on
our FTP site as
ftp://dev.ajubasolutions.com/pub/tcl/httpd/tclhttpd32-snapshot.zip
ftp://dev.ajubasolutions.com/pub/tcl/httpd/tclhttpd3.2-snapshot.tar.gz
again - there are various glitches in install/make that I'll iron out
in the next few days for an "official 3.2" release.

The snapshot is there now.

>>>Ted Dunning said:
 > 
 > 
 > Brent,
 > 
 > My status is that I wrote a non-event based version of geturl to completely
 > avoid the issue.  This is bad since this version does not allow the event
 > loop to continue, but good since it works and the server being called is
 > pretty fast.  We can regain CPU utilization in our application since we have
 > a load balancing proxy in front of our servers anyway.
 > 
 > Do I understand that the recommended idiom for calling geturl inside a .tml
 > file is this?
 > 
 >      set sock $Url(sock)
 >      Httpd_Suspend $sock
 >      set t [http::geturl ...]
 >      Httpd_Resume $sock ?args?
 > 
 > Should I call Httpd_Suspend in my Url_ReadPost fix as well?
 > 
 > 
 > -----Original Message-----
 > From: Brent Welch [mailto:[EMAIL PROTECTED]]
 > Sent: Thursday, October 12, 2000 11:35 AM
 > To: Ted Dunning
 > Subject: Re: TclHttpd 3.2 
 > 
 > 
 > Hi - I'm still catching up on email and just found this.
 > What is your status?
 > If a domain implementation does "http::geturl" then you'll need to
 > use "Httpd_Suspend" to clear out file events first.  3.2 adds a
 > "Httpd_Resume" as well.
 > 
 > >>>Ted Dunning said:
 >  > 
 >  > Brent,
 >  > 
 >  > We are having a serious problem with hangs in tclhttpd.  I suspect that
 > it
 >  > is due to my calling another tclhttpd using the http package.  The
 > symptom
 >  > is a server hang with high CPU usage.  I suspect a lost or stolen
 > fileevent
 >  > (although it possibly just be mislaid).  One key symptom is that a single
 >  > threaded replay of transactions does not seem to cause the hang.
 >  > 
 >  > We are still running 3.0.3.
 >  > 
 >  > Do you think that either 3.1 or 3.2 would solve this issue?
 >  > 
 >  > (eeek.  this bug, of course, only showed up in force when the system went
 >  > live)
 >  > 
 >  > -----Original Message-----
 >  > From: Brent Welch [mailto:[EMAIL PROTECTED]]
 >  > Sent: Tuesday, September 26, 2000 8:56 PM
 >  > To: Steve Blinkhorn
 >  > Cc: [EMAIL PROTECTED]
 >  > Subject: Re: TclHttpd 3.2
 >  > 
 >  > 
 >  > Can you speak CVS?  The 3.2 sources are up-to-date there.
 >  > I'll put up a 3.2snapshot.tar.gz on our ftp site:
 >  > ftp://ftp.ajubasolutions.com/pub/tcl/httpd/tclhttpd3.2-snapshot.tar.gz
 >  > ftp://ftp.ajubasolutions.com/pub/tcl/httpd/tclhttpd32-snapshot.zip
 >  > 
 >  > I'll have time next week to put some QA into a real 3.2 release.
 >  > 
 >  > >>>Steve Blinkhorn said:
 >  >  > > Right - the .rc file isn't the place.
 >  >  > > In 3.1 your only route is to add code the httpdthread.tcl
 >  >  > > In 3.2 I've added a "custom code" directory where you can
 >  >  > > put code without modifying the server.
 >  >  > 
 >  >  > Excellent idea - available when?   We're rapidly getting close to
 >  >  > needing to go live.
 >  >  > 
 >  >  > -- 
 >  >  > Steve Blinkhorn <[EMAIL PROTECTED]>
 >  > 
 >  > --        Brent Welch     <[EMAIL PROTECTED]>
 >  >   http://www.ajubasolutions.com
 >  >   Scriptics changes to Ajuba Solutions
 >  >   scriptics.com => ajubasolutions.com
 >  > 
 > 
 > --   Brent Welch     <[EMAIL PROTECTED]>
 >      http://www.ajubasolutions.com
 >      Scriptics changes to Ajuba Solutions
 >      scriptics.com => ajubasolutions.com
 > 

--      Brent Welch     <[EMAIL PROTECTED]>
        http://www.ajubasolutions.com
        Scriptics changes to Ajuba Solutions
        scriptics.com => ajubasolutions.com


Reply via email to