Re: [ANNOUNCEMENT] Updated: bash-4.1.13-6
On 09/29/2014 04:52 PM, Buchbinder, Barry (NIH/NIAID) [E] wrote: >> These functions contain '-' in their name; that's a limitation of >> the downstream forked patch I applied early to get 4.1.13-6 out the >> door. Upstream solved it in a nicer manner, so you can once again have >> functions with '-' in the name. Please try again with the just-released >> 4.1.14-7. > > I understand that to mean that 4.1.14-7 will be able to export functions with > '-' in > their names. Correct; also, I already verified it worked locally for me to export a function with '-' when using 4.1.14-7. > >>> I was able to get rid of this by commenting out the functions (make-log, >>> configure-log, and prt-alias) and removing them from the list of functions >>> being >>> set and exported in ~/.bashrc. I rarely use these functions and would be >>> unlikely >>> to use them in a subshell so that works as a solution. >> >> Having the function wasn't the problem, it was exporting it. You can >> still have the function even in 4.1.13-6, as long as you don't export >> it. > > Changed to underscores. All is well. Also an option. :) -- Eric Blake eblake redhat com+1-919-301-3266 Libvirt virtualization library http://libvirt.org signature.asc Description: OpenPGP digital signature
RE: [ANNOUNCEMENT] Updated: bash-4.1.13-6
Eric Blake sent the following at Monday, September 29, 2014 5:29 PM >On 09/29/2014 03:23 PM, Buchbinder, Barry (NIH/NIAID) [E] wrote: >> Eric Blake (cygwin) sent the following at Saturday, September 27, 2014 2:14 >> AM >> >> Using the new version of bash, two scripts that I use have both started >> giving me >> the following error message(s). >> >> /usr/bin/bash: error importing function definition for `BASH_FUNC_make-log' >> /usr/bin/bash: error importing function definition for >> `BASH_FUNC_configure-log' >> /usr/bin/bash: error importing function definition for `BASH_FUNC_prt-alias' > >These functions contain '-' in their name; that's a limitation of >the downstream forked patch I applied early to get 4.1.13-6 out the >door. Upstream solved it in a nicer manner, so you can once again have >functions with '-' in the name. Please try again with the just-released >4.1.14-7. I understand that to mean that 4.1.14-7 will be able to export functions with '-' in their names. >> I was able to get rid of this by commenting out the functions (make-log, >> configure-log, and prt-alias) and removing them from the list of functions >> being >> set and exported in ~/.bashrc. I rarely use these functions and would be >> unlikely >> to use them in a subshell so that works as a solution. > >Having the function wasn't the problem, it was exporting it. You can >still have the function even in 4.1.13-6, as long as you don't export >it. Changed to underscores. All is well. >> Another thing happening is that some scripts are now sometimes giving this >> error >> message, but only when the problem function exports are not being exported. >> >> Segmentation fault (core dumped) > >Umm, that's not good. Can you please come up with a minimal reproduction >formula? Which program is dumping core? Bash, or something else? Is it >due to trying to execute the function that was not imported? It wasn't reproducible, but a while after I sent my email various windows stopped responding, I couldn't close them, the machine wouldn't shut down, and ended up unplugging the machine so I could reboot. Let's assume that something unrelated to cygwin was going on. >> Any hints about what I should look for to make export of my bash shell >> functions >> reliable? > >Upgrading to the next version :) Will do! Thanks for all your help. - Barry Disclaimer: Statements made herein are not made on behalf of NIAID.
Re: [ANNOUNCEMENT] Updated: bash-4.1.13-6
On 09/29/2014 03:23 PM, Buchbinder, Barry (NIH/NIAID) [E] wrote: > Eric Blake (cygwin) sent the following at Saturday, September 27, 2014 2:14 AM > > Using the new version of bash, two scripts that I use have both started > giving me > the following error message(s). > > /usr/bin/bash: error importing function definition for `BASH_FUNC_make-log' > /usr/bin/bash: error importing function definition for > `BASH_FUNC_configure-log' > /usr/bin/bash: error importing function definition for `BASH_FUNC_prt-alias' These functions contain '-' in their name; that's a limitation of the downstream forked patch I applied early to get 4.1.13-6 out the door. Upstream solved it in a nicer manner, so you can once again have functions with '-' in the name. Please try again with the just-released 4.1.14-7. > > I was able to get rid of this by commenting out the functions (make-log, > configure-log, and prt-alias) and removing them from the list of functions > being > set and exported in ~/.bashrc. I rarely use these functions and would be > unlikely > to use them in a subshell so that works as a solution. Having the function wasn't the problem, it was exporting it. You can still have the function even in 4.1.13-6, as long as you don't export it. > > Another thing happening is that some scripts are now sometimes giving this > error > message, but only when the problem function exports are not being exported. > > Segmentation fault (core dumped) Umm, that's not good. Can you please come up with a minimal reproduction formula? Which program is dumping core? Bash, or something else? Is it due to trying to execute the function that was not imported? > > Any hints about what I should look for to make export of my bash shell > functions > reliable? Upgrading to the next version :) -- Eric Blake eblake redhat com+1-919-301-3266 Libvirt virtualization library http://libvirt.org signature.asc Description: OpenPGP digital signature
RE: [ANNOUNCEMENT] Updated: bash-4.1.13-6
Eric Blake (cygwin) sent the following at Saturday, September 27, 2014 2:14 AM Using the new version of bash, two scripts that I use have both started giving me the following error message(s). /usr/bin/bash: error importing function definition for `BASH_FUNC_make-log' /usr/bin/bash: error importing function definition for `BASH_FUNC_configure-log' /usr/bin/bash: error importing function definition for `BASH_FUNC_prt-alias' I was able to get rid of this by commenting out the functions (make-log, configure-log, and prt-alias) and removing them from the list of functions being set and exported in ~/.bashrc. I rarely use these functions and would be unlikely to use them in a subshell so that works as a solution. What makes this strange is that I have other shell functions that are defined in ~/.bashrc and exported. They have not shown up in the error messages. As it happens, the problem functions are the last 3 on the export list. However, adding another function to be exported doesn't cause a new error message for just one function that is now at the end of the export list. Another thing happening is that some scripts are now sometimes giving this error message, but only when the problem function exports are not being exported. Segmentation fault (core dumped) Any hints about what I should look for to make export of my bash shell functions reliable? Thanks, - Barry Disclaimer: Statements made herein are not made on behalf of NIAID.
Re: [ANNOUNCEMENT] Updated: bash-4.1.13-6
Eric Blake writes: > On 09/29/2014 10:14 AM, Achim Gratz wrote: >> Eric Blake (cygwin) writes: >>> A new release of bash, 4.1.13-6, has been uploaded and will soon reach a >>> mirror near you; leaving the previous version at 4.12-5. >> >> Just out of curiosity, why is this release version -6 instead of -1? > > Because of how upstream bash numbers things. [...] Thanks. Regards, Achim. -- +<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+ SD adaptations for Waldorf Q V3.00R3 and Q+ V3.54R2: http://Synth.Stromeko.net/Downloads.html#WaldorfSDada -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Re: [ANNOUNCEMENT] Updated: bash-4.1.13-6
On 09/29/2014 10:14 AM, Achim Gratz wrote: > Eric Blake (cygwin) writes: >> A new release of bash, 4.1.13-6, has been uploaded and will soon reach a >> mirror near you; leaving the previous version at 4.12-5. > > Just out of curiosity, why is this release version -6 instead of -1? Because of how upstream bash numbers things. They only ship a 4.1 tarball, and then 13 official patches that must be applied by the maintainer on top of that 4.1 tarball. However, it modifies the micro version number according to which upstream patches have been applied. So, this is my sixth time building bash 4.1 for cygwin, and it's just that bash's micro number bumped from 4.1.12 to 4.1.13 in between my -5 and -6 builds. When I eventually get bash 4.3 out the door, it will start at 4.3.27-1 (or a higher micro, as there is still a lot of churn in upstream bash to now patch lower-priority bugs that are no longer Shell Shock attacks thanks to patch 27, but which are still local crashers). -- Eric Blake eblake redhat com+1-919-301-3266 Libvirt virtualization library http://libvirt.org signature.asc Description: OpenPGP digital signature
Re: [ANNOUNCEMENT] Updated: bash-4.1.13-6
Eric Blake (cygwin) writes: > A new release of bash, 4.1.13-6, has been uploaded and will soon reach a > mirror near you; leaving the previous version at 4.12-5. Just out of curiosity, why is this release version -6 instead of -1? Regards, Achim. -- +<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+ Wavetables for the Waldorf Blofeld: http://Synth.Stromeko.net/Downloads.html#BlofeldUserWavetables -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Re: [ANNOUNCEMENT] Updated: bash-4.1.13-6
On 2014-09-29 09:18, Buchbinder, Barry (NIH/NIAID) [E] wrote: Eric Blake (cygwin) sent the following at Saturday, September 27, 2014 2:14 AM I also turned on the (undocumented) 'bash --wordexp' mode (actually, that happened in 4.1.12-5, although I failed to mention it at the time), which allows the C library call wordexp() to now function. From a user standpoint, why would one want to use --wordexp? What is the effect of turning on this mode? It's not meant to be used by users. It exists solely to be used by libc (cygwin) as an implementation of wordexp(3). Yaakov -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
RE: [ANNOUNCEMENT] Updated: bash-4.1.13-6
Eric Blake (cygwin) sent the following at Saturday, September 27, 2014 2:14 AM >I also turned on the (undocumented) 'bash --wordexp' mode (actually, >that happened in 4.1.12-5, although I failed to mention it at the time), >which allows the C library call wordexp() to now function. From a user standpoint, why would one want to use --wordexp? What is the effect of turning on this mode? Thanks, - Barry Disclaimer: Statements made herein are not made on behalf of NIAID.