Thanks...

I was also getting curious about this mechanism, is there any weak points
you can spot, because if not, I'm thinkin of using this for bugtracking in
ALL future projects. Especially those, which will have updates, because the
mass of data this could produce is astounding. Just think: reliable
quantitative bugreports!

Pekka


> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On
> Behalf Of Penworks Corporation
> Sent: 9. kesakuuta 2001 3:21
> To: [EMAIL PROTECTED]
> Cc: [EMAIL PROTECTED]
> Subject: RE: <lingo-l> need advice on alerthook
>
>
>
>
> Well, if you're not in a position to eradicate all bugs that
> could create a
> dialog before shipping, but you've tried, then certainly doing something
> like you're suggesting makes a nice insurance.
>
> Both myself and Bruce just wanted to make the point that you
> shouldn't use
> alertHook in lieu of bug repair.  It sounds like you're not that sort of
> person, but believe me - there are some people out there that would view
> that as preferably to fixing them.
>
> Still, if you've done your best, but want to provide just in case there's
> something you've missed, then trapping and logging them and doing
> the other
> things would sounds like you're trying to do the best you can under the
> conditions to ship, so good luck, and hope you have an opportunity to get
> the last few ones out real soon!
>
> - Tab
>
>
> At 02:26 AM 6/9/01 +0300, Pekka Buttler wrote:
> >Tab Julius wrote:
> >
> > > Well, it'll help mask errors, but if you have an error, now
> you'll never
> > > know where it was or what the user was doing.
> > >
> > > All you're doing is suppressing the error MESSAGE, but
> whatever the error
> > > (non-existent variable, undefined handler, non-list, etc),
> the error will
> > > still exist and will probably raise its ugly head shortly
> thereafter in
> > > some hard-to-track-down side-effect.
> >
> >and Bruce Epstein wrote:
> >
> > > IMNSHO it is insane to ship a project with a known bug by
> disabling the
> > > error message. How can any of *us* predict whether it will affect the
> > > project when we don't know what is causing the error or the
> nature of your
> > > project?
> >
> >OK guys. Thanks for the advice. The thing is this:
> >
> >I've been working for 6 days now at 18-20 hours per day behind
> my machine,
> >and I can be pretty sure we are doing all we can about this.
> >The problem with runtime errors is, they are not really useful. The user
> >get's annoyed and clicks [quit] in a situation where there's a
> small error
> >which would possibly not even have any impact. And is further annoyed by
> >seeing an error message at all...
> >
> >But you have a point, what comes to tracking down bugs, it gets
> really hard
> >without the error messages unless...
> >
> >I think I've come up with a compromise / tool.
> >1# I take Bruce's advice and write that nice ini-line
> >2# I suppress error messages
> >3# I log them
> >4# I upload them
> >
> >on alerthook me, err, msg, type
> >   global datestring, errorlog, lastErrorTimes
> >--datestring =  hh:mm:ss mm/dd/yyyy
> >-- errorlog 2 dimensional array of errors
> >-- lastErrorTimes = list containing 3 last error timestamps
> >   quitOnEnd = 0
> >   if the runmode = "projector" then
> >     if type = #script then
> >       LogLine = []
> >       errorLoc = the frame&"/"&the moviename
> >       add LogLine, datestring
> >       add LogLine, errorLoc
> >       add LogLine, err
> >       add LogLine, msg
> >       add errorlog, LogLine
> >       curErrorTime = the milliseconds
> >       if lasterrortimes[1] + 1000 > curErrorTimes then
> >        -- if we get here the error is unrecoverable (loop)
> >         quitOnEnd = 1
> >       else
> >       add lastErrorTimes, curErrorTime
> >       delete lasterrorTimes[1]
> >       Writelog(errorLog) -- converts the array to a csv - table
> and saves it
> >       if quitOnEnd = 0 then
> >         return 1 --(invisibly dismisses the error alert and continues)
> >       else
> >         return 0 --(invisibly dismisses the error alert and quits=
> >       end if
> >
> >     end if
> >   end if
> >end
> >
> >--
> >Now if this works, and a user has a problem, I get more detailed
> info then I
> >would (on average get from a 'dumb user' AND i not only get the
> one bug, but
> >I get all of them)
> >
> >Now, if we combine this with a feature, that if #internetconnected = TRUE
> >the log is posted to a cgi script (invisibly offcause <GRIN>).
> The old log
> >is deleted etc....
> >
> >What do you think? Should I go and catch some sleep, or am I
> onto something?
> >
> >Pekka
> >
> >
> >
> >
> >
> >[To remove yourself from this list, or to change to digest mode, go to
> >http://www.penworks.com/LUJ/lingo-l.cgi  To post messages to the list,
> >email [EMAIL PROTECTED]  (Problems, email [EMAIL PROTECTED])
> >Lingo-L is for learning and helping with programming Lingo.  Thanks!]
>
>
> [To remove yourself from this list, or to change to digest mode, go to
> http://www.penworks.com/LUJ/lingo-l.cgi  To post messages to the list,
> email [EMAIL PROTECTED]  (Problems, email [EMAIL PROTECTED])
> Lingo-L is for learning and helping with programming Lingo.  Thanks!]
>


[To remove yourself from this list, or to change to digest mode, go to
http://www.penworks.com/LUJ/lingo-l.cgi  To post messages to the list,
email [EMAIL PROTECTED]  (Problems, email [EMAIL PROTECTED])
Lingo-L is for learning and helping with programming Lingo.  Thanks!]

Reply via email to