[Rose-public] Missing EDG binary

Lin, Pei-Hung lin32 at llnl.gov
Fri Jan 22 14:46:14 PST 2016


Hi,
I have checked the server but could not find any EDG binary associated
with signature ³86034d03f84022814d56f2e36d00666eb7a8f812².
Could you pull from rose-develop repository and try again?  That should
try to use the latest EDG binary generated 2 days ago.

Thanks,
Pei-Hung


On 1/22/16, 14:08, "rose-public-bounces at nersc.gov on behalf of Jan-Patrick
Lehr" <rose-public-bounces at nersc.gov on behalf of
j_lehr at stud.tu-darmstadt.de> wrote:

>Hi,
>
>as far as I can tell (1) is a bug in the rose installation scripts which
>should be fixed and seems to be fixed in the development snapshot
>located at https://github.com/rose-compiler/rose-develop/
>
>I can't answer whether (2) is a general issue or if it is by design. The
>idea behind the checksum generation certainly is to provide you with a
>binary version of the EDG frontend which is guaranteed (maybe call it
>likely) to work with the compiler you are using. I guess that this is
>necessary to handle ABI compatibility issues which arise with different
>c++ compiler versions. However, I don't know, why gcc 4.8.5 should be
>different from gcc 4.8.1 as it should be solely bug fixes and completely
>ABI stable. Maybe someone from the Rose team can answer that question?
>
>Cheers
>JP
>
>On 1/22/2016 10:55 PM, Tim wrote:
>> Hello Jan,
>>
>> do you think this should be used as a work around? I am not
>> sure if this is a problem with our local setup or a general
>> problem in the rose installation scripts.
>>
>> Cheers!
>>
>> Tim
>>
>> On 01/21/2016 11:21 PM, Jan-Patrick Lehr wrote:
>>> Hey everyone,
>>>
>>> this somewhat looks like a problem we had, which was solved in two
>>> steps:
>>>
>>> 1) make sure you fix the typo in
>>> src/frontend/CxxFrontend/Makefile.am:107
>>>     - There is a 'g' missing in the domain ending
>>>
>>> 2) When we were using gcc 4.8.5 it seems that the edg bin signature
>>> differes from the one for gcc 4.8.1, what is why we build using
>>> make EDGBIN_NTRIES=5
>>>
>>> Honestly, I do not completely remember whether we used EDGBIN_NTRIES=5
>>> or a set it to a higher number, but that is probably the option you
>>>want
>>> to look for.
>>>
>>> Best,
>>> JP
>>>
>>> On 22.01.2016 00:59, Tim wrote:
>>>> Hello,
>>>>
>>>> I have some issues with building rose as well.
>>>> In detail, the edg file is missing.
>>>> Can somebody help me out what is wrong?
>>>>
>>>> cheers!
>>>>
>>>> Tim
>>>>
>>>>
>>>> make[4]: Entering directory
>>>> `/local/lecs/schmidtt/rose-develop/objects/src/frontend/CxxFrontend'
>>>> Making all in Clang
>>>> make[5]: Entering directory
>>>> 
>>>>`/local/lecs/schmidtt/rose-develop/objects/src/frontend/CxxFrontend/Cla
>>>>ng' 
>>>>
>>>> make[5]: Nothing to be done for `all'.
>>>> make[5]: Leaving directory
>>>> 
>>>>`/local/lecs/schmidtt/rose-develop/objects/src/frontend/CxxFrontend/Cla
>>>>ng' 
>>>>
>>>> make[5]: Entering directory
>>>> `/local/lecs/schmidtt/rose-develop/objects/src/frontend/CxxFrontend'
>>>> checking for EDG (SHA1)....
>>>> checking for EDG (MD5)....
>>>>
>>>> *** error finding EDG binary version
>>>> edg-generate-sig: could not find an exact version of the EDG binary.
>>>>    EDG is the C/C++ frontend from the Edison Design Group and is not
>>>> distributed
>>>>    as part of the ROSE source code. However, the ROSE team routinely
>>>> compiles
>>>>    EDG and makes binary versions available for download and linking
>>>> with
>>>> ROSE.
>>>>    The binary version of EDG must be compatible with the version of
>>>> ROSE
>>>> with
>>>>    which it is being linked, and we use hashes to compute the EDG
>>>> binary
>>>> version
>>>>    number from its ROSE dependecies with
>>>> $ROSE/scripts/edg-generate-sig.
>>>>
>>>>    We could not find a version of EDG that is an exact match
>>>>    for your version of ROSE.  If you're feeling lucky, you can try
>>>> extending
>>>>    the search and choosing a close match, but there is no guarantee
>>>> that
>>>>    the chosen binary EDG will be compatible with your version of
>>>> ROSE--be
>>>>    ready for strange behavior, core dumps, etc.  You can perform an
>>>> extended
>>>>    search by using the --ntries=N switch for this script
>>>> (edg-generate-sig),
>>>>    or running "make EDGBIN_NTRIES=N".
>>>>
>>>>    The following information is relevant:
>>>>      Time          = Wed Jan 20 17:39:44 PST 2016
>>>>      ROSE src dir  = /local/lala/rose-develop
>>>>      nTries        = 1
>>>>      Commit        = 32e21de281f6079620fa2fbfc39e3faa0f78f868 HEAD
>>>>      Manifest file = 8825740c8ae1e9b5c0621b0c32fe388e8b7a6506
>>>>      Num versions  = 5515
>>>>      Need (SHA1)   = 86034d03f84022814d56f2e36d00666eb7a8f812
>>>>      Need (MD5)    = 1fcc87778bb67a6c8d25a6e0593920c7
>>>>
>>>> EDG.tar.gz not found, checking if
>>>> 
>>>>roseBinaryEDG-4-7-x86_64-pc-linux-gnu-GNU-4.4-86034d03f84022814d56f2e36
>>>>d00666eb7a8f812.tar.gz
>>>>
>>>> is available
>>>> 
>>>>roseBinaryEDG-4-7-x86_64-pc-linux-gnu-GNU-4.4-86034d03f84022814d56f2e36
>>>>d00666eb7a8f812.tar.gz
>>>>
>>>> not found, download it!
>>>> Destination not reachable: download EDG binary from the following link
>>>> when the network is available:
>>>> 
>>>>http://www.rosecompiler.org/edg_binaries/roseBinaryEDG-4-7-x86_64-pc-li
>>>>nux-gnu-GNU-4.4-86034d03f84022814d56f2e36d00666eb7a8f812.tar.gz
>>>>
>>>>
>>>> Symbolic link
>>>> 
>>>>roseBinaryEDG-4-7-x86_64-pc-linux-gnu-GNU-4.4-86034d03f84022814d56f2e36
>>>>d00666eb7a8f812.tar.gz
>>>>
>>>> to EDG.tar.gz!
>>>> GZIP= gunzip -c EDG.tar.gz |tar -xf -
>>>>
>>>> gzip: EDG.tar.gz: unexpected end of file
>>>> tar: This does not look like a tar archive
>>>> tar: Exiting with failure status due to previous errors
>>>>
>>>> _______________________________________________
>>>> rose-public mailing list
>>>> rose-public at nersc.gov
>>
>
>-- 
>Jan -Patrick Lehr
>
>eMail: j_lehr at stud.tu-darmstadt.de
>
>_______________________________________________
>rose-public mailing list
>rose-public at nersc.gov



More information about the rose-public mailing list