1) I think I've fixed the first issue in a recent commit. 2) Parrot has a relatively small number of built-in native call frames or "thunks". If you want the ability to build others you have a few options: You can set up a file of NCI thunk signatures for Parrot to read and add. Otherwise you can install libffi and reconfigure/rebuild parrot with libffi support. I can walk you through these things on IRC.
--Andrew Whitworth On Sat, Jun 2, 2012 at 12:48 PM, Bart Wiegmans <[email protected]> wrote: > Hello everybody, > > Working on mod_parrot, I find myself with two problems I cannot (yet) solve. > The first is that using compreg('anything') from within the embedded > interpreter causes a segfault because the compreg_hash is null. I need > this to work to run a HLL compiler and external scripts. I have no > idea where it is set or if this represents a genuine a bug. > > The second is that it turns out the symbols for apache are available > at runtime, i.e. I can (could) use NCI to call apache functions > directly. This works to a point, because NCI will not call anything > with even slightly complex arguments (for instance, 2 PMCs together > with void result, does not work.). The error is 'No NCI thunk for > signature'. I would be much helped if somebody could explain why it > doesn't, and if there is a way to fix this. > > Kind regards, > Bart Wiegmans > > 2012/6/2, [email protected] > <[email protected]>: >> Send parrot-dev mailing list submissions to >> [email protected] >> >> To subscribe or unsubscribe via the World Wide Web, visit >> http://lists.parrot.org/mailman/listinfo/parrot-dev >> or, via email, send a message with subject or body 'help' to >> [email protected] >> >> You can reach the person managing the list at >> [email protected] >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of parrot-dev digest..." >> >> >> Today's Topics: >> >> 1. config/auto/sizes.pm: Unused code (James E Keenan) >> 2. Re: config/auto/sizes.pm: Unused code (Gerhard R.) >> 3. Re: config/auto/sizes.pm: Unused code (Andy Dougherty) >> 4. Re: config/auto/sizes.pm: Unused code (James E Keenan) >> 5. Re: config/auto/sizes.pm: Unused code (James E Keenan) >> 6. whiteknight/io_cleanup1 branch needs feedback (and help!) >> (Andrew Whitworth) >> 7. Re: config/auto/sizes.pm: Unused code (Andy Dougherty) >> 8. Re: config/auto/sizes.pm: Unused code (James E Keenan) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Fri, 01 Jun 2012 12:31:14 -0400 >> From: James E Keenan <[email protected]> >> To: [email protected] >> Subject: config/auto/sizes.pm: Unused code >> Message-ID: <[email protected]> >> Content-Type: text/plain; charset=ISO-8859-1; format=flowed >> >> Friends, >> >> Today when I performed a 'git pull' I noticed that there had been >> updates to config/auto/sizes.pm and t/steps/auto/sizes.t. I believe >> these updates were written several months ago, mostly by Gerhard R., but >> only recently merged into master. I ran coverage analysis which you can >> view starting here: >> http://thenceforward.net/parrot/coverage/configure-build/coverage.html >> >> Examining the coverage report for auto::sizes >> (http://thenceforward.net/parrot/coverage/configure-build/config-auto-sizes-pm.html), >> >> I noticed that internal subroutine _find_type_ge() was completely >> uncovered. This meant that not only was there nothing in the test file >> invoking it -- there was nothing in the module invoking it either! >> Examination of config/auto/sizes.pm confirmed that the subroutine is >> defined but never used. >> >> In general we should not retain code in the configuration system that we >> are not currently invoking. Is there any reason why we should retain >> this subroutine? If not, then I will remove it as part of looking at >> the module to see if there is any other still-uncovered code we can >> write tests for. >> >> Thank you very much. >> Jim Keenan >> >> >> >> ------------------------------ >> >> Message: 2 >> Date: Fri, 1 Jun 2012 21:02:13 +0200 >> From: "Gerhard R." <[email protected]> >> To: [email protected] >> Subject: Re: config/auto/sizes.pm: Unused code >> Message-ID: >> <CABH=7HYur2F2cLbAecn2BP4P_2CoD6q7QpLOPVy=mwypefy...@mail.gmail.com> >> Content-Type: text/plain; charset=ISO-8859-1 >> >> Hi, and sorry for replying out-of-thread. >> >>> I noticed that internal subroutine _find_type_ge() was completely >>> uncovered. This meant that not only was there nothing in the test file >>> invoking it -- there was nothing in the module invoking it either! >> >> That's a result of commit https://github.com/parrot/parrot/commit/762dddd5 >> >>> Is there any reason why we should retain this subroutine? >> >> No. I just failed to remove the sub when replacing its single invocation >> with a call to _find_type_min_ge(). >> >>> If not, then I will remove it as part of looking at the module to see >>> if there is any other still-uncovered code we can write tests for. >> >> Thanks for picking up my slack - much appreciated. >> >> Gerhard R. >> >> >> ------------------------------ >> >> Message: 3 >> Date: Fri, 1 Jun 2012 15:49:02 -0400 (EDT) >> From: Andy Dougherty <[email protected]> >> To: [email protected] >> Subject: Re: config/auto/sizes.pm: Unused code >> Message-ID: >> <[email protected]> >> Content-Type: TEXT/PLAIN; charset=US-ASCII >> >> On Fri, 1 Jun 2012, James E Keenan wrote: >> >>> Friends, >>> >>> Today when I performed a 'git pull' I noticed that there had been updates >>> to >>> config/auto/sizes.pm and t/steps/auto/sizes.t. I believe these updates >>> were >>> written several months ago, mostly by Gerhard R., but only recently >>> merged >>> into master. >> >> It looks like github issue #642 was inadvertantly closed with this merger. >> >> It should be reopened. The sample command line given in that issue: >> >> perl Configure.pl --intval="long long int" --opcode="long long int" >> >> still fails to produce a working parrot on a system where >> sizeof(long long) > sizeof(void *). >> >> -- >> Andy Dougherty [email protected] >> >> >> ------------------------------ >> >> Message: 4 >> Date: Fri, 01 Jun 2012 18:02:17 -0400 >> From: James E Keenan <[email protected]> >> To: [email protected] >> Subject: Re: config/auto/sizes.pm: Unused code >> Message-ID: <[email protected]> >> Content-Type: text/plain; charset=ISO-8859-1; format=flowed >> >> On 6/1/12 3:49 PM, Andy Dougherty wrote: >>> On Fri, 1 Jun 2012, James E Keenan wrote: >> >>> >>> It looks like github issue #642 was inadvertantly closed with this >>> merger. >>> It should be reopened. The sample command line given in that issue: >>> >>> perl Configure.pl --intval="long long int" --opcode="long long int" >>> >>> still fails to produce a working parrot on a system where >>> sizeof(long long)> sizeof(void *). >>> >> >> I have reopened issue #642. Let's move further discussion back into >> that issue. >> >> Thank you very much. >> Jim Keenan >> >> >> >> ------------------------------ >> >> Message: 5 >> Date: Fri, 01 Jun 2012 18:20:04 -0400 >> From: James E Keenan <[email protected]> >> To: [email protected] >> Subject: Re: config/auto/sizes.pm: Unused code >> Message-ID: <[email protected]> >> Content-Type: text/plain; charset=ISO-8859-1; format=flowed >> >> On 6/1/12 3:02 PM, Gerhard R. wrote: >>>> Is there any reason why we should retain this subroutine? >>> >>> No. I just failed to remove the sub when replacing its single invocation >>> with a call to _find_type_min_ge(). >>> >>>> If not, then I will remove it as part of looking at the module to see >>>> if there is any other still-uncovered code we can write tests for. >>> >>> Thanks for picking up my slack - much appreciated. >>> >> >> Thanks for your quick response. I have removed that sub in commit >> https://github.com/parrot/parrot/commit/db9fc1209d. >> >> I may do some other touch-ups as time permits to boost test coverage. >> >> Thank you very much. >> Jim Keenan >> >> >> >> ------------------------------ >> >> Message: 6 >> Date: Fri, 1 Jun 2012 20:34:12 -0400 >> From: Andrew Whitworth <[email protected]> >> To: [email protected] >> Subject: whiteknight/io_cleanup1 branch needs feedback (and help!) >> Message-ID: >> <CAEPDB4qu19UbM52my-=Mo2kWNTVqjh1gh7YF5FOq-0pCQ4z=h...@mail.gmail.com> >> Content-Type: text/plain; charset=ISO-8859-1 >> >> I've been working hard this week on the whiteknight/io_cleanup1 >> branch. This is essentially a huge rewrite of the IO subsystem to >> start fixing some of the largest problems with it. This is the first >> and the largest stage of a series of planned upgrades to IO. >> >> In the branch the build does not complete, though it does get most of >> the way through. The parrot binary, parrot-nqp and winxed all build. >> Coretest runs but there are several failures which I have not >> addressed yet. Some bits of functionality have not yet been >> implemented, or if implemented are not tested (seek, peek and tell >> operations on FileHandle, most operations on Pipes and sockets) >> >> There are two things I would like: First, feedback on the branch >> including the architecture, organization, algorithms, and other >> details. Second, I could really use some help debugging the build >> failures, fixing broken tests (especially codestd tests, which have >> been neglected so far) and testing. >> >> Thanks, >> >> --Andrew Whitworth >> >> >> ------------------------------ >> >> Message: 7 >> Date: Fri, 1 Jun 2012 22:12:45 -0400 (EDT) >> From: Andy Dougherty <[email protected]> >> To: James E Keenan <[email protected]> >> Cc: [email protected] >> Subject: Re: config/auto/sizes.pm: Unused code >> Message-ID: >> <[email protected]> >> Content-Type: TEXT/PLAIN; charset=US-ASCII >> >> On Fri, 1 Jun 2012, James E Keenan wrote: >> >>> On 6/1/12 3:49 PM, Andy Dougherty wrote: >>> > On Fri, 1 Jun 2012, James E Keenan wrote: >>> >>> > >>> > It looks like github issue #642 was inadvertantly closed with this >>> > merger. >>> > It should be reopened. The sample command line given in that issue: >>> > >>> > perl Configure.pl --intval="long long int" --opcode="long long >>> > int" >>> > >>> > still fails to produce a working parrot on a system where >>> > sizeof(long long)> sizeof(void *). >>> > >>> >>> I have reopened issue #642. Let's move further discussion back into that >>> issue. >> >> Thank you. I think all the details needed are in the original ticket (or >> even in the title of the ticket). Unfortunately, posting on github >> requires a github account, which I don't have and don't want to have >> to set up just to report occasional problems or fixes, so I'll just let it >> go from here. >> >> -- >> Andy Dougherty [email protected] >> >> >> >> ------------------------------ >> >> Message: 8 >> Date: Fri, 01 Jun 2012 22:17:40 -0400 >> From: James E Keenan <[email protected]> >> To: [email protected] >> Subject: Re: config/auto/sizes.pm: Unused code >> Message-ID: <[email protected]> >> Content-Type: text/plain; charset=ISO-8859-1; format=flowed >> >> On 6/1/12 3:02 PM, Gerhard R. wrote: >>>> Is there any reason why we should retain this subroutine? >>> >>> No. I just failed to remove the sub when replacing its single invocation >>> with a call to _find_type_min_ge(). >>> >>>> If not, then I will remove it as part of looking at the module to see >>>> if there is any other still-uncovered code we can write tests for. >>> >>> Thanks for picking up my slack - much appreciated. >>> >> >> Thanks for your quick response. I have removed that sub in commit >> db9fc1209ddbdaeb43ee559e06892c4e64517156 >> >> I may do some other touch-ups as time permits to boost test coverage. >> >> Thank you very much. >> Jim Keenan >> >> >> >> ------------------------------ >> >> _______________________________________________ >> http://lists.parrot.org/mailman/listinfo/parrot-dev >> >> >> End of parrot-dev Digest, Vol 46, Issue 1 >> ***************************************** >> > _______________________________________________ > http://lists.parrot.org/mailman/listinfo/parrot-dev _______________________________________________ http://lists.parrot.org/mailman/listinfo/parrot-dev
