> since the new server isn't standing here in my office i can not test if older elog 
> versions don't have the problem too. but if your idea of stack or buffer overflow 
> is right, i guess that hoping of the helpfull ms pach or a update of the compiler 
> will be the only what we could do  :-(  it's always the same using software fom a 
> company named "kleinstweich"...  ;-( 

;-))) like that one...

I tried to reduce the memory consumption in the threaded display, can you please try 
the modified version under

http://midas.psi.ch/elog/download/windows/elogd.exe

I must have misunderstood what you meant. That aside, I don't know how to compile 
the elogd.c so any help would be appreciated. I would like to learn how to also.
Thanks.

> > I even upgraded to the 2.5.1 version. Then I placed the elogd.c in the SRC 
> > folder. Also placed the images in the themes folder.
> 
> As I wrote, you have to get the latest elogd.c, revision 1.272 and *COMPILE* it 
> (or wait for the next official release). If you cannot compile elogd.c, let me 
> know and I can send you the executable.

BUG#1
Given:

Attributes = Category, Status, Started, Finished, Activity, Headline
Type Started = date
Type Finished = date
Options Status = 1-To Do{1}, 2-Opened{2}, 3-Closed{3}, 4-Suspended{4}
{2} Preset Started = $date
{3} Preset Finished = $date
{4} Preset Finished = $date

There seems to be no setting of Date format (or Time format?) that gives a
meaningful value to attributes Started or Finished.  Even "Date format = %s"
- which gives number of seconds since 1970 - gives bad results.

BUG#2
In single entry view, elog returns an unformatted date value for "Type =
date" attributes - even when it has been manually entered with the date picker.

> I agree. It's most probably a stack overflow problem. The stack size must be 
less  
> on a win 2003 server compared to a xp machine. I know that the threaded display  
> requires some more stack space than the other displays. But unfortunately I have  
> no win 2003 server here to try. 
 
since the new server isn't standing here in my office i can not test if older elog 
versions don't have the problem too. but if your idea of stack or buffer overflow 
is right, i guess that hoping of the helpfull ms pach or a update of the compiler 
will be the only what we could do  :-(  it's always the same using software fom a 
company named "kleinstweich"...  ;-( 
 
is there a posibility in elog to switch off the threaded view or to say what kind 
of views are allowed? this would be helpfull in this case as well as it may be  a 
nice feature for other users too. elog is very powerfull, fo some users i think 
too powerfull with to many possibilities. i think about something like you made in 
the menu / guest menu option or maybe a flag to say "don't" make threaded dispaly 
possible", like e-mail formating or something else? 
 
 

> the problem only ocures on win 2003 server. I d'dn't test it under xp, 2000, 
> 2002 beause i don't have any of this systems. but it looks that no one else 
> has this problem. would be helpfull to know if there anybody out in 
> cyberspase using elog 2.5.1 with win 2000 or 2002 or even better with win 
> 2003 server? or do you have a 2003 server at psi? 
> for me it looks like a incompatibility of a stack, boffer overrun or somethin 
> in the memory.

I agree. It's most probably a stack overflow problem. The stack size must be less 
on a win 2003 server compared to a xp machine. I know that the threaded display 
requires some more stack space than the other displays. But unfortunately I have 
no win 2003 server here to try.

hello stefan 
 
sorry for the delay, was out of the office today. i had tested the same config 
and logbook under a "pseudo" win nt installation (suse linux 8.2 with vmware 
4.01 (virtaual pc simualtion) and a nativ win nt 4.0 sp6a: there is no 
problem! as well as this configuration works under linux as daemon. 
 
the problem only ocures on win 2003 server. I d'dn't test it under xp, 2000, 
2002 beause i don't have any of this systems. but it looks that no one else 
has this problem. would be helpfull to know if there anybody out in 
cyberspase using elog 2.5.1 with win 2000 or 2002 or even better with win 
2003 server? or do you have a 2003 server at psi? 
for me it looks like a incompatibility of a stack, boffer overrun or somethin 
in the memory. for me it really only occurs under win 2003. thats what i'm 
wondering. the system memory test of the server was ok (no virus, non 
controller problem or something, else, checked the hole stuff agian, just to 
be really shure!) , other apps don't have any problmes (it isonly iis 6 and 
mercure mail server runnig on this system with about 20gb free disk and about 
500mb ram totally and a pentium 4). 
 
if you think you have a chance to reproduce the problem i can zip the config 
file and the log for you. but in my opinion, you will need a win 2k3.  
 
i'd chdcked the file system rights again to be shure that they are ok. also 
i'd checked that elog run as a system service under win 2k3. it should have 
all rights. changing to run eleog with administrator rights didn't change 
anything. 
 
would be great to find out what's happen. in the mean time i let it untreaded, 
hoping no one will click treaded view...  ;-)  for security, the servie is 
configured to restart if it hangs and to boot the system in an emergency, but 
it shouldn't happen evry hour... 
 
thakX! 
ueli 
 
 
 
> All I can offer is send me your config file for the problematic logbook, and  
> I will try to reproduce the error. 
 
 

> This was a bug which only occurs under certain conditions, that's why nobody 
> has seen it yet. I fixed it, the new version is available under CVS:
> 
> http://midas.psi.ch/cgi-bin/cvsweb/elog/src/elogd.c?rev=1.275

Great! 

I'm inclined to wait until you release a new rpm with the changes, as I'm 
likely to mucky things up by trying to upgrade from the cvs archive rather than 
using the rpm mechanics. 

But glad to hear it's fixed.  

By the way, ELOG is a fantastic tool. I'm very thankful for all the effort gone 
into it. I do wish there were standard sql hooks, as I'm a bit fearful of how 
it will scale. But the front-end, which is often most neglected, is well 
thought out.  Thanks so much, Val

This was a bug which only occurs under certain conditions, that's why nobody 
has seen it yet. I fixed it, the new version is available under CVS:

http://midas.psi.ch/cgi-bin/cvsweb/elog/src/elogd.c?rev=1.275