Menu
▾
▴
Re: [Bashdb-devel] bashdb 4.4-0.92: 5 tests fail
|
From: Rocky B. <roc...@gm...> - 2016-12-03 21:37:23
|
One last little thing. Just because you *think *you have configured not to compile files needing bash source, that doesn't necessarily mean that was the case. That's why details are important. It definitely looks like the configuration was set to try to build source. And the fact that read.c didn't barf in finding bash headers (but instead elsewhere) supports that belief. At the end of the configure script we dump what's up. For example: Location : /bin/bash Prefix : /usr/share prefix checks out We will try to build the set0 builtin to set variable $0 using source located at /src/build/bash-4.4. That import output was omitted in your log. But even if the "Will try to" line was missing since the configure run wasn't clean that too may be suspect. So the next thing I would look at is config.status and for the configuration above we would have S["BASH_SRC"]="/src/build/bash-4.4" But independent of any of this the fact that read.c was compiled means that the configuration was most likely somehow with some sort of bash source set. remake -x on a run will often also elucidate what's happening in a build. On Sat, Dec 3, 2016 at 3:24 PM, Rocky Bernstein <roc...@gm...> wrote: > Ok. Looked at this more. As I said, bash 4.4 changes the API > vaid_array_reference. As best as I can tell, it just breaks the old API > without adding anything. > > As for the ac_aux_dir problems, I get that too now. That too seems to be a > difference in how autoconf and automake work over the years and what they > now expect. > > Commit 68d03c3a85252447e04a87a686c7b14ae18616d should address both of > these. > > On Sat, Dec 3, 2016 at 11:42 AM, jean-christophe manciot < > act...@gm...> wrote: > >> Here are the details you've asked for: >> bash --version >> GNU bash, version 4.4.0(1)-release >> on Ubuntu 16.10 4.8 >> >> You have missed the fact that the configure line with --with-bash-src=... >> has been *commented out* because of the bug described in the comment >> located just above it and not reported. It has no effect in this bug >> report. >> >> The log & bug reported by this mail is the consequence of the non >> commented configure & make lines. >> >> On Sat, Dec 3, 2016 at 3:48 PM, Rocky Bernstein < >> roc...@gm...> wrote: >> >>> Jean-Christophe - >>> >>> Thanks for reporting the problems you encountered. I have a little >>> trouble reading the report because, although there is a bit that is >>> specific, various oher details aren't there. >>> >>> So here's my take on what the log given. >>> >>> 'readc.c:848:46: error: too few arguments to function >>>> ‘valid_array_reference’' >>> >>> >>> is probably caused by a change in the bash API to the function >>> valid_array_reference(). >>> The offending line in bashdb is: >>> >>> if (legal_identifier (varname) == 0 && valid_array_reference >>> (varname) == 0) >>> >>> What version of bash did you run against? If you or someone eise in the >>> devel lis is up for it, perhaps you can figure how to change read.c for the >>> changed var_array_reference call. >>> >>> If you remove --with-bash-src=/home/actionmystique/Program-Files/ >>> Ubuntu/Bash/git-bash from the configure option, bashdb won't try to >>> compile read.c. >>> >>> That particular function that is getting compiled allows bashdb to do a >>> command-completion kind of read inside it debugger read loop. It not needed >>> for overall basic functioning, but I guess it is a nice thing to have. >>> Perhaps bash now offers a way to handle completion inside its builtin >>> read() function. If that's the case we should remove compiling this. >>> Perhaps you or someone can check? >>> >>> As for setting $ac_aux_dir to '.', that may work for you in your >>> environment but it is symptomatic of some sort of configure script failure. >>> Any shell variable that starts ac_ is something the configure script sets >>> (ac stands I guess for autoconf). So if this is set wrong or is not set, >>> then things are probably misconfigured. When bashdb is misconfigured, the >>> test scripts may fail. >>> >>> >>> On Fri, Dec 2, 2016 at 2:25 PM, jean-christophe manciot < >>> act...@gm...> wrote: >>> >>>> Hi there, >>>> >>>> The following script run with branch=master & tag=release-4.4-0.92: >>>> >>>> echo -------- >>>> echo Cleaning >>>> echo -------- >>>> cd git-bashdb >>>> sudo -u actionmystique -H git-reset-clean-pull-checkout.sh >>>> $branch >>>> $tag >>>> >>>> echo ------------------ >>>> echo Building configure >>>> echo ------------------ >>>> export NOCONFIGURE=yes >>>> sudo -Eu actionmystique -H ./autogen.sh >>>> >>>> echo ------------------------------------------------------- >>>> echo Working around a bug related to $ac_aux_dir/config.sub >>>> echo ------------------------------------------------------- >>>> # Cf. https://sourceforge.net/p/bashdb/bugs/42/ >>>> sudo -u actionmystique -H sed -i 's|$ac_aux_dir|.|g' ./configure >>>> >>>> echo ----------- >>>> echo Configuring >>>> echo ----------- >>>> # Bugs 'readc.c:848:46: error: too few arguments to function >>>> ‘valid_array_reference’' >>>> # sudo -u actionmystique -H ./configure >>>> --with-bash-src=/home/actionmystique/Program-Files/Ubuntu/Bash/git-bash >>>> \ >>>> sudo -u actionmystique -H ./configure --prefix=/usr/share >>>> --sysconfdir=/etc --localstatedir=/var >>>> >>>> echo --------- >>>> echo Compiling >>>> echo --------- >>>> sudo -u actionmystique -H make >>>> >>>> echo -------- >>>> echo Checking >>>> echo -------- >>>> sudo -u actionmystique -H make check >>>> >>>> leads to *5 tests errors: attached test-suite.log* >>>> >>>> The whole build log is available on request. >>>> -- >>>> Jean-Christophe >>>> >>>> ------------------------------------------------------------ >>>> ------------------ >>>> Check out the vibrant tech community on one of the world's most >>>> engaging tech sites, SlashDot.org! http://sdm.link/slashdot >>>> _______________________________________________ >>>> Bashdb-devel mailing list >>>> Bas...@li... >>>> https://lists.sourceforge.net/lists/listinfo/bashdb-devel >>>> >>>> >>> >> >> >> -- >> Jean-Christophe >> > > |
