PDA

View Full Version : dfGUIv3 problem..



GOLDENBALLSAINTYORK
06-28-2002, 07:25 PM
This seems to work fairly well on my PC..
however sometimes It says I'm doing 2050 structures a minute..and the rest of the display panel is likewise erroneous!
I don't trust the upload, client stop features..but on manually stopping with 'Q' and uploading..then resstarting the GUI interface it starts and runs as normal..
I've deleted it in the meantime for ver.1.4

Athlon 1009 MHz..384 Mb Ram..Win. XP Home NTFS ????:rolleyes:

Digital Parasite
06-28-2002, 09:19 PM
I need more information if you want me to help you.

How long has dfGUI been running when this happens? Are you able to reproduce the problem? If yes, what do you do? Have you changed the date/time on your machine recently?

What do you mean you don't trust the client upload/stop features? What isn't working properly with it?

For the benchmarking information it looks at the current time and progress.txt file. If you change your date/time it will screw up the benchmarking. If you have just started dfGUI it takes a while for the benchmark info to be accurate. Do you have the # structures properly entered in dfGUI because that will screw up the benchmarking info as well?

Jeff.

stappel
06-29-2002, 04:15 AM
I have the same problem. It mostly happens after the client and gui is running for a longer time (>> 3 days). The best I can see it at is at struct per day. it goes from 35.000 to 158.000 a day.

- no time/date changes on my machine
- upload could be the problem. there is usualy more then 1 file buffered.

Stanley

Digital Parasite
06-29-2002, 07:44 AM
stappel: are you running the DF client in offline mode or online mode. Do you manually upload the results (by hitting the UPLOAD button). Do you launch the DF client from dfGUI either visible or hidden or are you running it as a service? What OS are you running on?

Jeff.

Kris
06-29-2002, 10:35 AM
Same observation. W2K, no services, no manual uploads requested, about 12 hours dfGUI uptime, full-time connection, no time changes, several files buffered for upload, I don't recall how I launched df client, the only interesting info to contribute is that it happened to one instance only on a dual board, and both were started identically, I believe. The other instance continued to perform as usual. I noticed it from a vnc window initially, if that would be a clue... I love the program and I hope this helps some. Thanks for your work.

Digital Parasite
06-29-2002, 10:43 AM
Thanks everyone for the info. It seems that most people have several files buffered for upload (probably due to the server acting strangely) so that may be a clue (although I have several files buffered myself and not experiencing this problem). I am going away for the long weekend so I will take a look at it next week to see if I can track down the problem.

Jeff.

Digital Parasite
06-29-2002, 11:56 AM
Hi everyone,

Talking with a few other people it seems their dfGUI benchmarks went haywire when the DF server started acting up on Friday afternoon. Even Dyyryath's benchmark script was reporting very strange results so it could have been the DF client itself putting strange information in the progress.txt file.

Now that the server is back up and running, restart your dfGUI program and let me know if the benchmark data screws up again. So far I haven't seen this problem.

Did anyone's dfGUI benchmark get screwed up before Friday? I haven't changed the code at all since very early versions so it has been doing the same thing for benchmarking for a long time now.

Jeff.

stappel
06-29-2002, 06:23 PM
I've seen the problem now a few times. I cannt confirm it just when the DF client is having problems uploading. I'll watch it and report when it happens again. I saw the it first time with the 0.6 version of the DFgui.

Digital Parasite
07-03-2002, 07:36 PM
Ok everyone, I went through all the dfGUI benchmark code. I couldn't find any obvious bugs but I went and added some checks so that if the DF client's progress.txt file reset the number of structures or something strange it will reset the benchmarking so you get correct info.

I could see a problem if you stopped the client and restarted it again (not using dfGUI) faster than the update time dfGUI was set for so dfGUI itself didn't notice the client had stopped and restarted. That could have thrown off the benchmark info.

I have released v1.4 which will hopefully solve the benchmark problem. If anyone still has problems with benchmark code going crazy with v1.4 please let me know and try to get all the details about what you were doing, etc... If you could grab the current progress.txt file and post it too that would be handy too.

Jeff.