[Boost.Test] Looking for Workaround: Subprocesses exiting with 0 reported as fatal error
Hi, Using Boost.Test version 1.35.0, we have run into the following bug http://svn.boost.org/trac/boost/ticket/1723 Is there an easy way to work around or fix this problem in Boost 1.35.0 (such as a patch), so we are not forced to upgrade to the SVN version of Boost.Test? Unfortunately, not starting a sub process is not an option, because it is not our software that starts the sub process but some additional library that we use. In case we have to upgrade to the SVN version of Boost, is there a recent SVN revision or tag that you would consider stable and recommend for use? Thanks in advance for any help. best regards, Florian
Florian Winter
In case we have to upgrade to the SVN version of Boost, is there a recent SVN revision or tag that you would consider stable and recommend for use?
There were very minimal bug fixes only (including one above). You should not have any problems upgrading to the svn version Gennadiy
On Sat, Jun 21, 2008 at 02:09:36AM +0000, Gennadiy Rozental wrote:
Florian Winter
writes: In case we have to upgrade to the SVN version of Boost, is there a recent SVN revision or tag that you would consider stable and recommend for use?
There were very minimal bug fixes only (including one above). You should not have any problems upgrading to the svn version
Is there any reason why you don't comment this in the bug report? Currently it looks like you never dealed with this bug. There is no comment in it, no reference to a revision which fixes it, no patch ... Jens
Jens Seidel
On Sat, Jun 21, 2008 at 02:09:36AM +0000, Gennadiy Rozental wrote:
Florian Winter
writes: In case we have to upgrade to the SVN version of Boost, is there a recent SVN revision or tag that you would consider stable and recommend for use?
There were very minimal bug fixes only (including one above). You should not have any problems upgrading to the svn version
Is there any reason why you don't comment this in the bug report? Currently it looks like you never dealed with this bug. There is no comment in it, no reference to a revision which fixes it, no patch ...
Not sure I understood you completely, but if you mean I should have closed bug report - I thought I did. If not I'll do. Gennadiy
On Sun, Jun 22, 2008 at 03:57:26PM +0000, Gennadiy Rozental wrote:
Jens Seidel
writes: On Sat, Jun 21, 2008 at 02:09:36AM +0000, Gennadiy Rozental wrote:
Florian Winter
writes: In case we have to upgrade to the SVN version of Boost, is there a recent SVN revision or tag that you would consider stable and recommend for use?
There were very minimal bug fixes only (including one above). You should not have any problems upgrading to the svn version
Is there any reason why you don't comment this in the bug report? Currently it looks like you never dealed with this bug. There is no comment in it, no reference to a revision which fixes it, no patch ...
Not sure I understood you completely, but if you mean I should have closed bug report - I thought I did. If not I'll do.
Yep, I mean you should close it. If such a closing only results in a minor flag change which may be missed by people (I missed it the last tme I looked at it!) it would be nice to see also a human readable comment (such as "is now fixed in trunk"). Trac provides the possibility to refer to commits and opens a diff viewer once clicked an a revision. It should not be difficult for you to refer to such a patch/revision in svn for trac. Strange, I do not find the Trac bug database anymore on the website. It's not below Support, Development, ... Ah, now I see it's a toplevel item in the right strip on the website. Once a subcategory (Welcome, Introduction, ...) is open it is missing (I use the KDE konqueror web browser, 3.5.5)! Mhm, lost it again!?? How to view bug report? Please improve the website! Jens
Jens Seidel
Trac provides the possibility to refer to commits and opens a diff viewer once clicked an a revision. It should not be difficult for you to refer to such a patch/revision in svn for trac.
Can you point me to the instructions how to do this? Gennadiy
On Mon, Jun 23, 2008 at 05:38:50PM +0000, Gennadiy Rozental wrote:
Jens Seidel
writes: Trac provides the possibility to refer to commits and opens a diff viewer once clicked an a revision. It should not be difficult for you to refer to such a patch/revision in svn for trac.
Can you point me to the instructions how to do this?
Yep, http://svn.boost.org/trac/boost/wiki/TicketWorkflow contains: Subversion tip: when you commit a change to Subversion that fixes ticket number NNN, include the text "Fixes #NNN" in your commit log. Trac will automatically close the ticket and cross-reference the commit with the ticket. I'm not sure whether this works if you later change the svn:log property. Jens
-----Ursprüngliche Nachricht----- Von: boost-users-bounces@lists.boost.org [mailto:boost-users- bounces@lists.boost.org] Gesendet: Sonntag, 22. Juni 2008 19:35 An: boost-users@lists.boost.org Betreff: Re: [Boost-users] [Boost.Test] Looking for Workaround: Subprocesses exiting with 0 reported as fatal error
Jens Seidel
writes: On Sat, Jun 21, 2008 at 02:09:36AM +0000, Gennadiy Rozental wrote:
Florian Winter
writes: In case we have to upgrade to the SVN version of Boost, is there a recent SVN revision or tag that you would consider stable and
recommend
for use?
There were very minimal bug fixes only (including one above). You should not have any problems upgrading to the svn version
Is there any reason why you don't comment this in the bug report? Currently it looks like you never dealed with this bug. There is no comment in it, no reference to a revision which fixes it, no patch ...
Not sure I understood you completely, but if you mean I should have closed bug report - I thought I did. If not I'll do.
This bug does not seem to be fixed in Boost 1.36.0. I still obtain errors when child processes exit with return code 0. Using the test code in trac ticket #1723 (http://svn.boost.org/trac/boost/ticket/1723): Test setup error: child has exited; pid: 503; uid: 26975; exit value: 0 Regards, Peter.
Peter Klotz writes:
This bug does not seem to be fixed in Boost 1.36.0. I still obtain errors when child processes exit with return code 0.
Using the test code in trac ticket #1723 (http://svn.boost.org/trac/boost/ticket/1723):
Test setup error: child has exited; pid: 503; uid: 26975; exit value: 0
I was able to get my unit tests working by setting the environment variable: BOOST_TEST_CATCH_SYSTEM_ERRORS=no It's a rather big hammer, but it got me going. Cheers, -- Robert
participants (5)
-
Florian Winter -
Gennadiy Rozental -
Jens Seidel -
Peter Klotz -
Robert Mecklenburg