2016 Beta versions ready to test!

Marco,

Strange, there are a bunch of things missing from the form in the screen shot! (Captions, boxes, etc.). Does the form always look like that, or are things missing just after the error occurs?

Tom

Those things are just missing, after the error occured. Without that error the form looks normal and intact. Example of this morning, after a restart of wxsimlite, caused by the error.


OK, a couple of follow-ups:

Gary, if that wret error happened right after a WXSIM run (when wret is called to make some files), it sounds like somehow the latest.wxf file got corrupted. I’d certainly like to investigate that; if I can find the problem and the cause, I can do something to prevent it in the future. Next time you see it, just send me the latest.wxf file and I can try it here.

Tornadochaser, OK, at least WXSIM-Lite can make a forecast, so I’m guessing the error happened during the analysis run (?). Unfortunately, the things missing in the screen shot keep me from seeing more about what it could be. Can you watch it attempt the analysis run and then describe just what it’s doing (what date it’s on, and whether it’s on the first or second pass) if and when the error occurs again? Also, does the error happen regardless of whether you’ve checked the box for using year-old data?

Thanks!

Tom

Hi Tom,

the strange thing is, that the analysis run works like a charm. After every crash, I restart WXSIM-Lite an initiate a analysis run via scheduler (run immediately). No errors. Like I wrote in my post from Monday, it seems that the error occurs while processing a forecast. I thought in the first place, that maybe a logfile of mine is corrupt, but autolearn is performing quite well and doesn

Hi Tom

I have exactly the same error as Tornadochaser with the error 62, with half the info on the wxsimlite screen is missing. I have it ticked to include last years information. Running wxsimlite independently shows no errors but it always seems to happen on the first run of the day. I am running the beta6 version. When this happens no forecast is produced.

regards

Dave

HI Tom

I’ve been getting the same error 62 the last few days. When it happens I go in an delete the “gfsdat” file that it hangs on, but the next day the error reappears and hangs on a different file. Good news is that WxSim does complete the forecast. I’m using Beta6 and it worked fine for the first week or 2.

Cheers

MikeyM


OK, I’m trying to solve these issues. It might help if I explain what changes I made in beta6, and generally how some things work, so maybe I or someone else will figure something out!

The routine which forecasts (both in the scan and in the forecast for the current day) now has a check to see if it’s reached the end of a gfsdat0… file, BEFORE reading the (missing) line and triggering an error. Instead of a program-stopping error, a red-texted “non-critical error message”, identifying the incomplete file, gets placed at the bottom of the form. I’ve tested this pretty well and it seems reliable.

So, the error 62 (input past end of file) shouldn’t be due to a short gfsdat file (unless it’s less than about two lines). I’m trying to think of what other files could trigger this. One might be the log file (like 32016lg.txt), especially if the message says that was the last file attempted to be opened. If this is the case, I’d like to see that log file. Conceivably, it might be a more “internal” file, like cormod0.txt, if it got improperly created. Basically, please look closely at the error message, and send me whatever file it said it opened last. Actually, if it says some file was the last one opened successfully, and then you got input past end of file, there is very likely something wrong with THAT file. I’d need to see it to (1) fix the file and/or (2) make WXSIM-Lite smart enough to recognize such an error in advance, and work around it, and/or really pinpoint the cause.

Another experiment to try is NOT using the last year data, which also tends to narrow things down. I have a feeling that some of these errors have surfaced because of some problem with old data, which went undetected before because that data wasn’t being used until now.

Thanks for your patience. I really want to get this all right - and then release this new version officially! :slight_smile:

Tom

I just thought of something that could cause problems like error 62, especially if it’s happening only on the first run of the day (or self-corrects after a manual run):

IF you are using either WeatherLink or VWS, WXSIMATE has to make a run (even without a WXSIM run after) BEFORE WXSIM-Lite. The reason for this is that WXSIMLite (and wret, for that matter) in these cases actually reads secondary files produced from the originals (like .wlk files or dbase.csv files), and this has to be done before wret or WXSIM-Lite go looking for the files. It shouldn’t be an issue with WD or Cumulus.

If you’re seeing the errors, please let me know which software you’re using!

Tom

Hi Tom

I am using WD .

I have unclicked ‘use last years data’ and see if it improves the situation with the 62 error

regards

Dave

HI Tom

I’m using WD and “Lite” ran fine overnight. That’s after I deleted the “gfsdat” files that “Lite” hung on yesterday. I’ll keep you updated. Thanks

Cheers

:smiley:

MikeyM

Perhaps it has been missed what Tom said about deleting gfsdat files a few posts back?

Tom,
On the 23 I too got an error 62 in Lite but at the same time (morning) there was the attached error. I found that Wxsimate was stuck halfway in downloading data. I manually started a download and all went fine. The next day there was no analysis or forecast in Lite and after restarting the Lite and running a manual analysis and forecast, I have had no more issues. Using WD. All good for now.


3-23.JPG

Solved (I think!) some users’ problems today. I just wrote to another user with the following information. I’m hoping this will help some of you!

I’ve spent the last couple of days solving (I think!) about three users’ (somewhat different) problems. I’ll share what I learned, and maybe something of it will take care of yours, too! I’m mostly talking WXSIM-Lite here, but some of it applies to autolearn/wret as well.

(1) One customer, who uses Weather Display, was getting errors in WXSIM-Lite, including “input past end of file”. It turned out that he really did have two or more corrupted (too short) gfsdat0… files. He was seeing the error in older versions and had not yet upgraded to 2016beta6. Once he did, the error went away (replaced by a non-intrusive message in red at the bottom of the form about the corrupted files). Deleting those files would then be an option to avoid even the little message at the bottom.

(2) Another customer was having errors with autolearn/wret. It turned out that he had an incorrect path listed in wret to his WD log files.

(3) A VWS user was getting WXSIM-Lite errors when trying to use the new feature for including year-old data. He did have gfsdat0… files going back far enough, but when I investigated further, it turned out that his VWS data, in dbase.csv, went back only to June, 2015.

An issue which may come up with both VWS and WeatherLink users is that, for both of these programs (but not Weather Display or Cumulus), WXSIMATE actually creates secondary files (like 32016wl.txt and 32016vws.txt) for wret and WXSIM-Lite to read. In normal use, these are created automatically. However, it may be that year-old data of this type doesn’t exist, because you weren’t running WXSIMATE (at least in the same way) back then. In this case, there’s the one-time task of creating the older files. To do so:

(1) Open WXSIMATE
(2) Go to Schedule and temporarily disable the timer
(3) On the right (green) side of the form, enter the year and month of an old file (like from 2015) you need to create
(4) In the boxes right after that, enter the last day of that month and then 23.99 for the time
(5) Click Import Data so it will create that month’s file
(6) Repeat for any months’ files which don’t yet exist

It’s really not necessary to switch to Cumulus from VWS, though you can if you want. You’re right that you’d have to wait a while for the learning routines to be helpful.

Tom - for both my last runs I’ve had the attached error. Running beta 6 but this is the same error as I had in post#79 on March 1st under beta4. The error message does not appear until just as the forecast completes. I can see no gfs download errors.

EDIT:…and on the third forecast all is back to normal, no errors.


wxsimfc.JPG

Hi Tony,

First, look at wdata.txt and see if it was made recently. Within it, look at the line like this:

Fcast_hr,prmsl,apcp,tcdc,wspd,wdir,hgt1000,hgt500,tmp950,tmp925,tmp900,tmp850,tmp800,tmp750,tmp700,rh950,rh925,rh900,rh850,rh800,rh750,rh700,rh500,rh300
2016-03-20-18

Note the date/time stamp on that GFS data. If it’s more than a day old, that’s what the program is talking about. You need to verify that WXSIMATE is running before each WXSIM forecast.

Let me know what you find!

Tom

Actually, also make sure that on the form under Import/Import Data in WXSIM, that you do have READY or GFS checked.

Tom

Hi Tom - thanks for that. I have checked everything as you suggested and it all looks okay. wdata.txt is up to date. If it happens again I now know which file to look at immediately afterwards…

However when I try and check WXSIM Import > Import Data I can’t get the form to appear. It just flashes and disappears. I do know that GFS was ticked though when I set up the program back in 2014. I’ve tried shutting WXSIM down and relaunching it but it’s still the same. In the morning I’ll shut down the PC and restart everything and see what happens.

Hi Tony,

That’s because auto run is activated (under Start/Auto run and other settings). Temporarily disable it and you will be able to access that form.

Tom

Thanks Tom - Ready or GFS was already ticked. I only had the error message on two occasions this month for a couple of runs each time. Everything has been fine since the last one.

Tom I had not updated to this latest beta 6 until this morning as I was away when you initially released the first beta and then I saw all these issues. So this morning I updated to beta 6 just prior to my first run of the day. I have now had 3 runs so far and all is going fine, no errors so far. Here’s hoping it stays that way.

Stuart