Windows Problems

[Brian the Fist]

As some of you have noted, you may be experiencing problems with the client update under Windows. No this is not a ploy to get everyone to swicth to Linux (although it WOULD make everything so much easier...).

We have 'upgraded' from the MSVC6 compiler we have been using in the past to the new Visual Studio .NET compiler. While all the code tested and worked fine under MSVC6, we foolishly assumed it would work under .NET as well (but we should know better by now ).

Thus as some of the most eager amongst you discovered, the initial Windows update had a problem causing it to crash after the 1st structure (error 0x000000c). Thus we quickly turned the server back off (of course). Looking at the cause of the problem, we found it was something to do with the optimization being done by the compiler. Given a choice between a client that crashes and an unoptimized one, we decided to quickly repackage the updates with optimization off. Thus it runs slower that normal but at least works.

Now we have had a bit more time to look at the error and have found a way around it with optimizations in place, although the reason for the crash itself is still largely unknown and is in a complex portion of the code in a 3rd party library.

Anyone who updated after 11AM EST today should have the fixed, full-speed version. The same applies if you downloaded from the web site after this time. For your convenience we have also made the EXE only available here:

ftp://ftp.mshri.on.ca/pub/distribfol...ldtrajlite.zip

simply unzip it and overwrite the 'bad' EXE (whichever one you have) and it should be fine. Note you can only update this way if you already have yesterday's client on your machine. If your machine has not updated yesterday/today, please update the normal way and you will get the working version now.

Similarly for the screensaver:

ftp://ftp.mshri.on.ca/pub/distribfol...trajlitess.zip

and unzip it over the old screensaver executable (foldtraj.scr) (search for it if unsure where it is, but itll be somewhere in your windows\system directory area)

We apologize for any inconvenience but this was a very unforeseen and strange problem which we didn't expect to arise from a simple upgrade in compilers.

On the bright side, we have fixed many bugs from the previous client. Please let us know of any new ones that should pop up so we can fix them ASAP. As noted in the whatsnew.txt, we have made a lot of (albeit minor) changes with this version so I won't be too surprised if there's one or two new errors we missed.

P.S. No one commented on the surprise funky download progress bar I added?
People only ever notice the bad stuff..

[pointwood]

P.S. No one commented on the surprise funky download progress bar I added?

My download speeds are too fast for me to notice

Thanks for responding so fast Howard!

/me thinks Howard owes a beer

[Wedge]

thanks for providing the fix so quickly

[Louis]

We love you man!

Originally posted by Brian the Fist
We apologize for any inconvenience but this was a very unforeseen and strange problem which we didn't expect to arise from a simple upgrade in compilers.

Unforeseen? We are talking about a micro$oft product, right? Well, I guess random complications are technically unforeseen...

[^7_of_9]

Originally posted by Brian the Fist
P.S. No one commented on the surprise funky download progress bar I added?
People only ever notice the bad stuff..

This is the first timme I'm going through a DF update myself as I only worked on DF I for a week or so, so I thought that was normal. But yes I did think that was cool though

P.S. I like DF II though for sure and plan on sticking with it. So far it's alot better than most the other projects I've seen and the way things were handled.

[willebenn]

The speed seems ok with todays update, but
it seems the benchmark is borked.
When running I see the following
opening rotamer Lib
Predicticting Sec. ...
[NULL_Caption] FATAL ERROR[065.012] input
AUTH_List.names.std.e.level.
0 is invalid value for level
Hit return

Just something for your to do list

[DocWardo]

Howard, the funky download bar was cool. I run almost all in service mode so I only saw it when I ran the update manually on one of my linux boxen.

just a question: anyone else notice that there was a 5 MB size difference between the two executables?

[Wedge]

Originally posted by DocWardo
just a question: anyone else notice that there was a 5 MB size difference between the two executables?

that's because the other files weren't inculded in the zipped file, only the exe.

[Pinky and the Brain]

Downloading and installed the latest client (EXE is 2.9 meg Dated 7/23 10:48am)

This client is running the Gen 0 at FULL SPEED again....... However, EVERY one of my PCs on the newest client is failing after gen 0 with (or similar error regarding VAL file):

[NULL_Caption] FATAL ERROR: [000.000] File qqd4qid9_0_protein_0004687.val is corrupt; cannot continue - replace file and start again, or manually delete filelist.txt
Hit Return


HELP!!!!

[HaloJones]

Client won't go past gen 0

[PinHead]

When just copying the foldtrajlite.exe to clients that took the update:

Windows 98 and Windows 2000 with clients running in quiet mode, I am getting an occasional 8 bytes of hex output to the screen.

Win 98
-078D2E80-
-078D2E80-

Win 2000
-03988128-

They still seem to be running and nothing in the error log.

[tak22]

Downloaded the exe over a working copy, it finished gen 0, then this when starting gen 1:

[NULL_Caption] FATAL ERROR: [000.000] Maketrj reported an error, cannot contine

Is this fixable or do I have to wait for DF to be up from maintenance to get a new copy?

[DocWardo]

Originally posted by Wedge
that's because the other files weren't inculded in the zipped file, only the exe.

No, I'm talking when I copied the new executable over into my distribfold directory.

old: 07/22/2003 05:26p 7,368,704 foldtrajlite.exe

new: 07/23/2003 10:41 AM 3,043,328 foldtrajlite.exe

[PinHead]

Just got that fatal error also!

Here's the error log entry:

========================[ Jul 23, 2003 12:48 PM ]========================
ERROR: [003.001] {extaaseq.c, line 991} Unable to fetch Biostruc xxxxxxxx_2_protein_0004101_min.val
ERROR: [005.001] {vismaketrj.c, line 347} Unable to determine number of models
FATAL ERROR: [000.000] {foldtrajlite2.c, line 3768} Maketrj reported an error, cannot continue

[Brian the Fist]

Originally posted by Pinky and the Brain
Downloading and installed the latest client (EXE is 2.9 meg Dated 7/23 10:48am)

This client is running the Gen 0 at FULL SPEED again....... However, EVERY one of my PCs on the newest client is failing after gen 0 with (or similar error regarding VAL file):

[NULL_Caption] FATAL ERROR: [000.000] File qqd4qid9_0_protein_0004687.val is corrupt; cannot continue - replace file and start again, or manually delete filelist.txt
Hit Return


HELP!!!!

Ugh. This is related to the same problem that was causing the crash. Methinks we need to return to MSVC6 immediately. I'm not up to gen 1 yet (how does everyone get there so fast??!??) but if I get the same error here I'll rebuild everything, AGAIN, on MSVC6 and it should be fine

[HaloJones]

Originally posted by Brian the Fist
Ugh. This is related to the same problem that was causing the crash. Methinks we need to return to MSVC6 immediately. I'm not up to gen 1 yet (how does everyone get there so fast??!??)

Because the first ten thousand structures have been running since yesterday!!!

[pfb]

getting similar problems - it'll do gen 0 then stop and won't go any further

took around 1.5 hours for gen 0 to run...

[Richard Clyne]

Words can not express my feelings about DF at this moment.

I could do with a "Go and get rat arsed, go to directly to jail, do not pass go, and do not collect £200" smillie at this moment.

[HaloJones]

Isn't this time to regress to the last client, bugs and all?

[CodeMonkey]

Do these things (displaying on a -qt window, BTW) mean anything?

-037D6E60-
-0382EF08-
-038C43C8-

[HansArne]

Using -qt -rt

[NULL_Caption] FATAL ERROR: [000.000] File qqd4qid9_0_protein_0004687.val is corrupt; cannot continue - replace file and start again, or manually delete filelist.txt
Hit Return

[Alien952]

There is still something wrong.......dunno

lost 2 x near 10,000 proteins
here the error log text of both
using -qt -rt -p -2 on both


========================[ Jul 23, 2003 5:45 PM ]========================

========================[ Jul 23, 2003 5:45 PM ]========================
ERROR: [003.001] {extaaseq.c, line 991} Unable to fetch Biostruc xxxxxxxx_0_protein_0007523.val
FATAL ERROR: [000.000] {foldtrajlite2.c, line 3488} File xxxxxxxx_0_protein_0007523.val is corrupt; cannot continue - replace file and start again, or manually delete filelist.txt

--------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------------

========================[ Jul 23, 2003 7:50 PM ]========================
ERROR: [003.001] {extaaseq.c, line 991} Unable to fetch Biostruc xxxxxxxx_0_protein_0006451.val
FATAL ERROR: [000.000] {foldtrajlite2.c, line 3488} File xxxxxxxx_0_protein_0006451.val is corrupt; cannot continue - replace file and start again, or manually delete filelist.txt


S**T.....don't know what's going on

Alan

[Morphy375]

@Howard

Why wait until You got this error?

Should I stay at office all over the night to get the boxen fixed?

========================[ Jul 23, 2003 5:51 PM ]========================
ERROR: [000.000] {foldtrajlite2.c, line 2070} Wed Jul 23 17:53:54 2003
Unable to check server status
ERROR: [000.000] {foldtrajlite2.c, line 2070} Wed Jul 23 19:50:28 2003
Unable to check server status
ERROR: [003.001] {extaaseq.c, line 991} Unable to fetch Biostruc xxxxxxxx_3_protein_0008922.val
FATAL ERROR: [000.000] {foldtrajlite2.c, line 3488} File xxxxxxxx_3_protein_0008922.val is corrupt; cannot continue - replace file and start again, or manually delete filelist.txt

[HaloJones]

Howard, few people are going to want to run the unoptimized code. It's just a waste of electricity. Please, put the old code from last week back on the servers so that we can run last week's code without it trying to get this week's code.

This is a disaster which must be regressed until you have new code that will work as advertised. I will happily beta-test if that is the issue but right now, you are losing your audience rapidly!

[Brian the Fist]

It is clear MSVC .NET does not like that part of the code so we have reverted to MSVC6. Please divert any anger directly to .NET
All links mentioned above, and update patches, and official version on the download page have now been uploaded to the MSVC6 compiler binary, both screensaver and text client.

Thanks for your patience.

[gistech1978]

thanks for the quick resolution, howard.

[Paratima]

Howard: I've told my boss several times that we should NOT migrate to MSCV.NET. MSCV6, I said, works just fine.

Thanks for supporting my position!

[DocWardo]

take home message:

it shouldn't be called .NET it should be called .NOT!

I hold no anger toward the project or you howard, then again I only have 6 boxen to take care of

[Rbreb13]

I was getting those same errors. Testing the new client now. That was fast Howard GJ!

[rshepard]

Should the previous client be blown out and a fresh install done? I'm still seeing the same error.

[Aegion]

Originally posted by rshepard
Should the previous client be blown out and a fresh install done? I'm still seeing the same error.

A fresh install would definately be best and should solve these issues.

[Brian the Fist]

If you were unfortunate enough to get the error about the .val file being corrupt at the end of generation 0, you will have to delete filelist.txt to get it going again, yes. The file is, indeed, corrupt after all. The fixed client cannot un-corrupt it. But it will not produce any corrupt files itself anymore.

[Jammy]

Thanks for the head up Howard! But I saw Paratima's post before I saw yours, sorry. BTW >>> I downloaded the "newest" client and wow! The speeds make the speeds of the client I downloaded yesterday seem like a snails pace (which they were, lol). My XP 2600 will be finishing up generation "0" in a few minutes, and last night I calculated it as finishing in 2 1/2 days.

Also my XP 1900 has speeded up with its work and so has my 1.33 T-Bird.

Sorry . . .but I did not have to go to the website and download the file so I did not see your "funky" download bar, lol.

Thanks again!

Jammy

BTW >>> When I downloaded that new client yesterday I had no problems downloading it because my df client auto-updated itself. It was just waiting for the server(s) to clear up so work could be started.

BTW >>> what happened to the points from the last 10 total generations from all three of my boxes that I managed to upload yesterday "before" the new client started?

Originally posted by Brian the Fist


Anyone who updated after 11AM EST today should have the fixed, full-speed version. The same applies if you downloaded from the web site after this time. For your convenience we have also made the EXE only available here:

ftp://ftp.mshri.on.ca/pub/distribfol...ldtrajlite.zip

simply unzip it and overwrite the 'bad' EXE (whichever one you have) and it should be fine. Note you can only update this way if you already have yesterday's client on your machine. If your machine has not updated yesterday/today, please update the normal way and you will get the working version now.


P.S. No one commented on the surprise funky download progress bar I added?
People only ever notice the bad stuff..

[Morphy375]

Originally posted by Brian the Fist
If you were unfortunate enough to get the error about the .val file being corrupt at the end of generation 0, you will have to delete filelist.txt to get it going again, yes. The file is, indeed, corrupt after all. The fixed client cannot un-corrupt it. But it will not produce any corrupt files itself anymore.

It's a nice new game to update 15 Offlineboxen three times a day....

[spldart]

I coulda warned yah that .NET is teh EvIl

[pfb]

Originally posted by Brian the Fist

It is clear MSVC .NET does not like that part of the code so we have reverted to MSVC6. Please divert any anger directly to .NET
All links mentioned above, and update patches, and official version on the download page have now been uploaded to the MSVC6 compiler binary, both screensaver and text client.

Thanks for your patience.

Howard,

I downloaded the new version (modified 23/07/2003, 1:43PM) and put it into a clean install but still get the

========================[ Jul 23, 2003 8:29 PM ]========================
ERROR: [003.001] {extaaseq.c, line 990} Unable to fetch Biostruc <handle>_0_protein_0003180.val
FATAL ERROR: [000.000] {foldtrajlite2.c, line 3485} File <handle>_0_protein_0003180.val is corrupt; cannot continue - replace file and start again, or manually delete filelist.txt

errors...?

[DocWardo]

silly question: but when you did the clean install did you really do a clean install (only surviing files: service.cfg, handle.txt, autoupdate.cfg. ? or did the .val files stay behind too.

[Slo]

Most recent foldtrajlite.zip and still the same...

-07B28E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B29E80-
-07B43E80-
-07B2AE80-
[NULL_Caption] FATAL ERROR: [000.000] File t31ti7ma_0_protein_0008654.val
annot continue - replace file and start again, or manually delete filelist.
Hit Return


Slo...

[HansArne]

YESSSS!!!
It's alive.
I'm on gen 1

[pfb]

Odd - my P4 seems OK but the same version on my Dual AMD isn't...

Another gen 0 for them...