Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Solaris testing of C++ library

15 views
Skip to first unread message

nolo...@gmail.com

unread,
Jun 15, 2016, 7:30:28 PM6/15/16
to
Hi Everyone,

My apologies for this message My apologies if you also received this on comp.unix.solaris.

I'm part of a free/open source project called Crypto++ (http://www.cryptopp.com/). Crypto++ is a C++ class library of cryptographic schemes written by Wei Dai. I'm one of the library stewards/maintainers when Wei is out of pocket.

We are in the midst of our next release. We are concerned we may not have enough Solaris and Sun Studio test coverage. We have access to a DL360 running Solaris 11.3, but we don't have access to an UltraSPARC.

We would like to ask those with developer experience, a dev config and spare cycles to perform the following and report back with results.

git clone https://github.com/weidai11/cryptopp.git
cd cryptopp
./cryptest.sh

The script is Sun Studio aware, so it should "just work" out of the box.

Thank you in advance for any testing you can perform.


----------


cryptest.sh is our "test script from hell". It repeatedly builds the library under different configurations and options. At each build it runs the self test and validation suites.

You can see the script at https://github.com/weidai11/cryptopp/blob/master/cryptest.sh, and read more about it at http://www.cryptopp.com/wiki/Release_process#Test_Script.


----------


If all goes as expected, you should see similar to:

$ cat cryptest-result.txt | tail -14
************************************************
************************************************

Testing started: Wed Jun 15 10:08:28 EDT 2016
Testing finished: Wed Jun 15 13:03:57 EDT 2016

38 configurations tested

No failures detected

No warnings detected

************************************************
************************************************

We are interested in warnings because a clean compile is a security gate. We don't use toy warnings - we are aggressive about the warning levels and clearing any findings produced.
0 new messages