From elzacho at gmail.com Thu Jul 3 00:29:42 2008 From: elzacho at gmail.com (Zach) Date: Wed, 2 Jul 2008 18:29:42 -0600 Subject: [fetter-devel] Verrazano tests do not generate correct CFFI bindings? Message-ID: <95f1acd0807021729u553cf8f7h605f2c0297ec8e75@mail.gmail.com> Hello, I am new to verrazano, although I have been wanting to play with it for a while. After a bit of fiddling, I believe I was able to get verrazano running (there were some mysterious problems). However, upon trying the examples, I find that there are still some problems. If it matters: SBCL 1.0.17 on OS X, gccxml 0.9, vzn pulled from darcs on 7-1-2008. I tried building posix bindings (the only ones that come close to working for me) and I get this line in the file it produces, (cffi::defctype _-darwin-uuid-t :unsigned-char :count 16) Which is not what CFFI expects (it's lambda list is something like (name type &optional documentation)). If I ignore the warning at the top of the file and edit this line to look like, (cffi::defctype _-darwin-uuid-t :unsigned-char) Things do work at least to some degree. Not sure how well they are supposed to work. For example, the posix test has posix:o-rdonly in it. It seems that is the mode we are opening the file in, read only. I suppose this should be a numeric constant. However, that symbol is unbound in my crippled posix package. Here is what I have tried: Various odd problems were due to bugs in the dependencies (i.e. I was using the snapshot of alexandria off cliki (from 2007), which breaks verrazano) and using gccxml 0.5 seems to cause several mysterious problems which manifest as unbound slots later on (upgrading to 0.9 seemed to fix these issues). I have the CVS version of CFFI, (I downloaded a autogenerated tarball from 2008-6-15). Looking in the archives I saw that someone suggest pulling from loliviera's CFFI branch, but looking at the darcs tree I see that defctype is similarly defined. Does any one know where I went wrong? Is there a list of known issues I can look at? Is there any output you would like that will help you diagnose? Zach -------------- next part -------------- An HTML attachment was scrubbed... URL: From osei.poku at gmail.com Tue Jul 29 23:33:55 2008 From: osei.poku at gmail.com (Osei Poku) Date: Tue, 29 Jul 2008 19:33:55 -0400 Subject: [fetter-devel] parsing of extern Message-ID: Hi, I am just beginning to look at verrazano code. I tried to generate OpenGL headers based on the code from the generate- example-bindings file. So I spent a while investigating this.... it seems gccxml is createing this node... however, when the gccxml:extern slot is being set, it complains that it cannot set the slot because this slot's type is boolean but the string "1" is the value being set. Is there something, that changed in gccxml or a way to convert a "0"/"1" to a boolean NIL/T value for boolean types? gcc-xml version is 0.9.0 i pulled verrazano from CVS today. Osei From attila.lendvai at gmail.com Wed Jul 30 08:01:35 2008 From: attila.lendvai at gmail.com (Attila Lendvai) Date: Wed, 30 Jul 2008 10:01:35 +0200 Subject: [fetter-devel] parsing of extern In-Reply-To: References: Message-ID: > I tried to generate OpenGL headers based on the code from the > generate-example-bindings file. when possible, always send a test-case in such situations (or at least a backtrace), so that we can reproduce the error locally. it should work and generating the stock opengl binding works for me: VERRAZANO-USER> (generate-opengl-binding) ; *** Processing binding :OPENGL-CFFI-BINDINGS $ gccxml --version ; Standard output: GCC-XML version 0.9.0 $ gccxml -I/usr/include -fxml="//tmp/872vzntemp.xml" "//tmp/872vzntemp.cpp" $ gccxml -I/usr/include --preprocess -dDI "//tmp/872vzntemp.cpp" >"//tmp/872vzntemp.mac" ; No value VERRAZANO-USER> btw, flexml.lisp has the interesting part of the code, search for "boolean". -- attila From osei.poku at gmail.com Wed Jul 30 15:13:00 2008 From: osei.poku at gmail.com (Osei Poku) Date: Wed, 30 Jul 2008 11:13:00 -0400 Subject: [fetter-devel] parsing of extern In-Reply-To: References: Message-ID: <4E2AC2E1-0753-4706-AB4B-87FB6FFB005D@gmail.com> On Jul 30, 2008, at 4:01 AM, Attila Lendvai wrote: >> I tried to generate OpenGL headers based on the code from the >> generate-example-bindings file. > > when possible, always send a test-case in such situations (or at least > a backtrace), so that we can reproduce the error locally. it should > work and generating the stock opengl binding works for me: Sure.. attached... The first thing is the result when I run generate- opengl-binding as is... The backtrace is obtained by removing the handler-bind wrapper around the generate-binding call. -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: backtrace-opengl.txt URL: -------------- next part -------------- > > > VERRAZANO-USER> (generate-opengl-binding) > ; *** Processing binding :OPENGL-CFFI-BINDINGS > $ gccxml --version > ; Standard output: > GCC-XML version 0.9.0 > $ gccxml -I/usr/include -fxml="//tmp/872vzntemp.xml" "//tmp/ > 872vzntemp.cpp" > $ gccxml -I/usr/include --preprocess -dDI "//tmp/872vzntemp.cpp" >> "//tmp/872vzntemp.mac" > ; No value > VERRAZANO-USER> > > btw, flexml.lisp has the interesting part of the code, search for > "boolean". > > -- > attila