Message boards :
Sixtrack Application :
Shiny new executables! Could you try them?
Message board moderation
Author | Message |
---|---|
Send message Joined: 23 Jan 17 Posts: 29 Credit: 375,570 RAC: 0 |
Hi all, we've made some shiny new executables that will soon be pushed out (first to more testing with actual BOINC, then to actual BOINC): https://github.com/SixTrack/SixTrack/releases/tag/v4.6.5 These are execs for Linux (32- and 64-bit), Windows(32- and 64-bit), OS X (Darwin, 64-bit), NetBSD (64-bit) and FreeBSD(64-bit). They pass all tests on our systems; however if you have the chance to try and run them on yours, that would be fantastic - especially if you have an old (Win XP?) and/or actual 32-bit systems hanging around. The best way to try them would be to run it in a folder with the "Sixin.zip" from this folder: https://github.com/SixTrack/SixTrack/tree/master/SixTest/fma I.e. put the Sixin.zip (Don't uncompress it!) together with the Sixtrack binary in an empty folder, open a terminal in that folder, and run SixTrack. This tests produces quite a bit of output (over several files, so run it in a "fresh" folder), and should take about 30 minutes to run. The actual "terminal" output should go to the file "fort.6". After it finishes, take a look at the file "fma_sixtrack". This should be the same as the file "fma_sixtrack.canonical" in the github repository (possibly excluding the Windows line-endings if you ran it on a Windows machine). If you do this, please leave a post with how it went (success/failure), which OS/distro and CPU architecture. Especially if something crashes, please let us know! Kyrre Sjobak, SixTrack developer@CERN |
Send message Joined: 23 Jan 17 Posts: 29 Credit: 375,570 RAC: 0 |
MD5SUMs should be:
|
Send message Joined: 14 Jan 10 Posts: 1268 Credit: 8,421,616 RAC: 2,139 |
This was a very fast unsuccessful test on my 32bit Win10 machine Terminal output: C:\SixTrack>SixTrack_465_libarchive_cr_boinc_api_crlibm_fast_tilt_cmake_Windows_gfortran.exe_static_AMD64_32bit CRITICAL ERROR in read_archive(): When opening archive 'Sixin.zip', err=-30 CRITICAL ERROR in read_archive(): Unrecognized archive format Contents of file fort.93: SIXTRACR starts very first time The system is 64bit, but the Win-OS is 32bit. |
Send message Joined: 23 Jan 17 Posts: 29 Credit: 375,570 RAC: 0 |
Dear Crystal Pellet, thanks a lot for trying it, stuff like this ( executable compatibility for systems that I don't have access to) is exactly what I'm trying to catch here! At least the executable ran; however it seems that it could not read the Sixin.zip (it did find the file tough...). Hmm, strange? Just to double-check that we're running it in the same way: 1) create an empty folder "test" under C:\Users\USERNAME\ (where my username has no spaces in case that matters 2) Download the .exe and move it into that folder 3) Download the .zip by clicking the "Download" button here: https://github.com/SixTrack/SixTrack/blob/master/SixTest/fma/Sixin.zip 4) Move the .zip into that folder too - do NOT unpack it or alter it in any way. 5) Open a windows console (CMD) 6) cd test 7) SixTrack_465_libarchive_cr_boinc_api_crlibm_fast_tilt_cmake_Windows_gfortran.exe_static_AMD64_32bit (like you did) ... wait ... 8) more fma_sixtrack lots of stuff like this: https://www.dropbox.com/s/ireq9m1myd08rae/Screenshot%202017-02-13%2016.33.27.png?dl=0 9) more fort.6 lots of stuff like this: https://www.dropbox.com/s/uykioephzdfvnfy/Screenshot%202017-02-13%2016.33.45.png?dl=0 If you have access to an md5sum utility, the sum should read like this: eaa3679611085da29bea20137be232cf Sixin.zip |
Send message Joined: 23 Jan 17 Posts: 29 Credit: 375,570 RAC: 0 |
I'm pretty sure this error is due to a corrupt Sixin.zip; I was able to reproduce it only by puting "asdf" in the header of the .zip file... Could you try to redownload? If you open the .zip file with "more" from the console, it should look something like this: https://www.dropbox.com/s/re79xk9cbdmvhem/Screenshot%202017-02-13%2017.27.38.png?dl=0 |
Send message Joined: 14 Jan 10 Posts: 1268 Credit: 8,421,616 RAC: 2,139 |
I'm pretty sure this error is due to a corrupt Sixin.zip; I was able to reproduce it only by puting "asdf" in the header of the .zip file... Could you try to redownload? I redownloaded it. Strange is that the size increased 59.922 bytes to 60.951 bytes, but is still wrong and not working. Somehow it thinks it's multivolume. At least now it's working: In your first post you told us to fetch the sixin.zip from https://github.com/SixTrack/SixTrack/tree/master/SixTest/fma In your newer post you suddenly wrote https://github.com/SixTrack/SixTrack/blob/master/SixTest/fma and there I could fetch the right sized Sixin.zip sized 346.869 bytes, although they're landing to the same spot. The exe is still running right now and it could take a bit longer, cause it's running on a slow Intel Atom Z3735F together with 3 LHC-Theory VM's. |
Send message Joined: 23 Jan 17 Posts: 29 Credit: 375,570 RAC: 0 |
Great! I don't know why something was wrong, and I don't know what you mean with multivolume. However when I go to the link https://github.com/SixTrack/SixTrack/tree/master/SixTest/fma and click on the "Sixin.zip", I get to https://github.com/SixTrack/SixTrack/blob/master/SixTest/fma/Sixin.zip (i.e. "blob") where I can click on the "download" or "view raw", both which fetches the .zip. Maybe you downloaded the webpage or something? Anyway, great that it works, and sorry that my instructions were not 100% clear! 346869 bytes is correct, this is the same as I get on my dev machine: fma $ ll Sixin.zip -rw-rw-r--. 1 kyrsjo kyrsjo 346869 Feb 9 12:09 Sixin.zip Thanks again! And if anyone has a WinXP machine, that would be absolutely fantastic... The oldest I have is 2008 Server, which *should* be equivalent, but one never knows. Old OS X and old non-redhat Linux would also be nice... |
Send message Joined: 14 Jan 10 Posts: 1268 Credit: 8,421,616 RAC: 2,139 |
Maybe you downloaded the webpage or something? Yeah probably. I right clicked the link (https://github.com/SixTrack/SixTrack/blob/master/SixTest/fma/Sixin.zip) and told to save as ... For the fma_sixtrack.canonical there is no download option, so I've to copy/paste into Notepad++ to check against my result. Edit: BTW: My test is ready. The contents of fma_sixtrack is the same as fma_sixtrack.canonical, except there are 1623 characters more. The same number as the number of lines. Maybe an extra <CR> per line in Windows? |
Send message Joined: 2 May 07 Posts: 2071 Credit: 156,141,325 RAC: 105,216 |
Have a xp(x86)pro as a Guest-VM. 346869 bytes for Sixin.zip - ok. bcrypt.dll not found is the message after start of sixtrack in Console. |
Send message Joined: 6 Sep 08 Posts: 116 Credit: 10,927,002 RAC: 2,464 |
Running on native 32bit winXP sp3. gets the "cannot find bcrypt.dll" message, too. This dll isn't on the systen at all. EDIT. Googling bcrypt yields this: Looking into getting Casablanca (the C++ REST SDK) working on Windows XP – the main issue seems to be the use of Crypto API Next Generation (CNG) , otherwise known as the BCrypt.DLL APIs. These APIs are only available on Vista and higher, so finding a workaround (replacement) for these APIs would be the top priority if trying to find a solution to the missing XP support. From here:- https://supportxpdotcom.wordpress.com/2013/02/23/casablanca-c-rest-sdk-xp-support/ |
Send message Joined: 23 Jan 17 Posts: 29 Credit: 375,570 RAC: 0 |
Thanks a lot! I feared something like this on XP. The issue is that a library we use to interact with .zip files, "libarchive", links dynamically to bcrypt and a few other system libraries on Windows, and there seems to be no way to link statically with system libraries on Windows. However looking into the CMakelist for Libarchive https://github.com/libarchive/libarchive/blob/master/CMakeLists.txt, it seems that I can disable "Crypto New Generation" https://msdn.microsoft.com/en-us/library/windows/desktop/aa376210(v=vs.85).aspx support when compiling Libarchive and it should not try to link with BCrypt... And since we anyway don't need encrypted .zip's, disabling this feature should not be any problem (and if we would have needed the feature, I could probably get libarchive to use openssl on Windows also). I'll try make and publish a new set of Windows execs later today, without this "feature"! |
Send message Joined: 23 Jan 17 Posts: 29 Credit: 375,570 RAC: 0 |
New windows execs: https://www.dropbox.com/sh/o9pa3h06fca9ndq/AADmCsgMpY6lNZvgrj3y8JIUa?dl=0 I'm not 100% sure if they are free of the bcrypt dependency tough; I did try to disable it but LDD tells me something is still linking it. Could you please try? |
Send message Joined: 6 Sep 08 Posts: 116 Credit: 10,927,002 RAC: 2,464 |
Tried the 32 bit one on this host but still the same failure, I'm afraid. |
Send message Joined: 2 May 07 Posts: 2071 Credit: 156,141,325 RAC: 105,216 |
32-bit +1 |
Send message Joined: 6 Sep 08 Posts: 116 Credit: 10,927,002 RAC: 2,464 |
Dependency walker (2.2.6, 32bit also on XP) gives this:- Error: At least one required implicit or forwarded dependency was not found. Warning: At least one module has an unresolved import due to a missing export function in a delay-load dependent module. The error refers to bcrypt and the warning to mpr.dll.... I think. Found this:- Some versions of SHLWAPI.DLL (like the one on Windows XP) have a delay-load dependency on the function WNetRestoreConnectionA in MPR.DLL. Missing delay-load functions are not a problem as long as the calling DLL is prepared to handle the situation. Dependency Walker flags all potential problems as it cannot detect if an application intends to handle the issue. In the case of SHLWAPI.DLL, this is not an problem as it does not require WNetRestoreConnectionA to exist and handles the missing function at runtime. This warning can be ignored. See the "How to Interpret Warnings and Errors in Dependency Walker" section in help for more details. |
Send message Joined: 14 Jan 10 Posts: 1268 Credit: 8,421,616 RAC: 2,139 |
Win10 32bit: exe-v1 and exe-v2 fma_sixtrack outfile from both tests matches Also tested on a Win7 64bit machine with a corresponding result. |
Send message Joined: 10 Aug 11 Posts: 2 Credit: 7,666,181 RAC: 0 |
Tested successfully on a Debian 8.7 32-bit x86/i686 VMware virtual machine. The test took much less time than mentioned though. EDIT: real time consumed: 4 min 34 sec Regards, dda |
Send message Joined: 9 Dec 14 Posts: 202 Credit: 2,533,875 RAC: 0 |
Tested on Debian 9 (testing) 64 bit in QEMU/KVM. fma_sixtrack files are identical, but the same what dda mentioned: it took just a couple of minutes, not 30 min (EDIT: about 4min 15 sec). |
Send message Joined: 9 Dec 14 Posts: 202 Credit: 2,533,875 RAC: 0 |
Also tested on Debian 9.0 (testing) 32 bit in QEMU/KVM. same result as for the 64 bit version: fma_sixtrack files are identical, but the same what dda mentioned: it took just a couple of minutes, not 30 min |
Send message Joined: 2 May 07 Posts: 2071 Credit: 156,141,325 RAC: 105,216 |
OpenSuse-Leap_42.2(x64) - Linux: 4.4.46-11-default AMD FX-8370E Vishera Finished successful after 9 min. Same Size as fma_sixtrack.canonical |
©2024 CERN