Hello,
Any news or things to be done before pike 7.6 release ?
Thanks, /Xavier
Well, we have silently been working on a 7.6 release for the last few weeks, with e.g some fixes for the latest bundle versions yesterday. I've had reports on problems with generating the configure files from the CVS version, which is the only issue that currently blocks a release AFAIK. It would be nice if everyone took a fresh CVS-check out and made
make make doc make snapshot
and reported any problems (and solutions :) here.
It would also be helpful if people downloaded the latest generated build from http://pike.ida.liu.se/generated/pikefarm/packages/7.6/latest and reported any concerns.
It would also be helpful if people downloaded the latest generated build from http://pike.ida.liu.se/generated/pikefarm/packages/7.6/latest and reported any concerns.
Is there a difference from running the xenoclient?
No, but you might want to check that you code actually runs with it and not just that it builds. Things like external modules, monger etc are not automatically tested.
I just built 7.6-20051018-184907 on OS X 10.4.2 (gcc 3.3) and here are the results:
make - OK
make doc - OK but 35-40 warnings about resolving packages
make snapshot - dies with "checking ansi prototype capability... no, giving up..."
make verify - 1 error (src/test_resolve.pike terminates with bus error), but hasn't that been broken for a long time, at least on OS X?
It would also be helpful if people downloaded the latest generated build from http://pike.ida.liu.se/generated/pikefarm/packages/7.6/latest and reported any concerns.
Seems to compile well on FreeBSD 6.0-RC1 SMP... excepted the need for specifying threads library or the autoconf will blown itself and add -pthread and -lc_r thread library...
So with CONFIGUREARGS="--with-thread-library=-lpthread" this is working well
eg:
make => ok make doc => ok make snapshot => not ok (strange error with ansi thing as reported before.)
I didn't have time yet to try with Caudium, but this is on my todo...
/Xavier
Important note the nettle bundle don't work on a machine that doesn't have nettle installed :
config.status: linking /usr/home/kiwi/Desktop/Downloads/Pike-v7.6-snapshot/bundl es/nettle-1.13/x86/machine.m4 to machine.m4 config.status: executing dummy-dep-files commands config.status: executing nettle-stdint.h commands config.status: creating nettle-stdint.h : __NETTLE_STDINT_H Compiling bundle nettle in /usr/home/kiwi/Desktop/Downloads/Pike-v7.6-snapshot/b uild/freebsd-5.4-release-i386/bundles/nettle.build... "Makefile", line 348: Could not find aes.o.d aes-decrypt-table.o.d aes-decrypt.o .d aes-encrypt-table.o.d aes-encrypt.o.d aes-set-encrypt-key.o.d aes-set-decrypt -key.o.d aes-meta.o.d arcfour.o.d arcfour-crypt.o.d arcfour-meta.o.d arctwo.o.d arctwo-meta.o.d base16-encode.o.d base16-decode.o.d base16-meta.o.d base64-encod e.o.d base64-decode.o.d base64-meta.o.d cast128.o.d cast128-meta.o.d blowfish.o. d cbc.o.d ctr.o.d des.o.d des3.o.d des-compat.o.d hmac.o.d hmac-md5.o.d hmac-sha 1.o.d hmac-sha256.o.d knuth-lfib.o.d md2.o.d md2-meta.o.d md4.o.d md4-meta.o.d m d5.o.d md5-compat.o.d md5-meta.o.d sha1.o.d sha1-compress.o.d sha1-meta.o.d sha2 56.o.d sha256-meta.o.d serpent.o.d serpent-meta.o.d twofish.o.d twofish-meta.o.d yarrow256.o.d yarrow_key_event.o.d sexp.o.d sexp-format.o.d sexp-transport.o.d sexp-transport-format.o.d bignum.o.d bignum-random.o.d sexp2bignum.o.d pkcs1.o.d pkcs1-rsa-md5.o.d pkcs1-rsa-sha1.o.d rsa.o.d rsa-sign.o.d rsa-verify.o.d rsa-md 5-sign.o.d rsa-md5-verify.o.d rsa-sha1-sign.o.d rsa-sha1-verify.o.d rsa-encrypt. o.d rsa-decrypt.o.d rsa-keygen.o.d rsa-compat.o.d rsa2sexp.o.d sexp2rsa.o.d dsa. o.d dsa-sign.o.d dsa-verify.o.d dsa-keygen.o.d sexp2dsa.o.d pgp-encode.o.d rsa2o penpgp.o.d buffer.o.d buffer-init.o.d realloc.o.d nettle-internal.o.d aesdata.o. d desdata.o.d shadata.o.d aes.po.d aes-decrypt-table.po.d aes-decrypt.po.d aes-e ncrypt-table.po.d aes-encrypt.po.d aes-set-encrypt-key.po.d aes-set-decrypt-key. po.d aes-meta.po.d arcfour.po.d arcfour-crypt.po.d arcfour-meta.po.d arctwo.po.d arctwo-meta.po.d base16-encode.po.d base16-decode.po.d base16-meta.po.d base64- encode.po.d base64-decode.po.d base64-meta.po.d cast128.po.d cast128-meta.po.d b lowfish.po.d cbc.po.d ctr.po.d des.po.d des3.po.d des-compat.po.d hmac.po.d hmac -md5.po.d hmac-sha1.po.d hmac-sha256.po.d knuth-lfib.po.d md2.po.d md2-meta.po.d md4.po.d md4-meta.po.d md5.po.d md5-compat.po.d md5-meta.po.d sha1.po.d sha1-co mpress.po.d sha1-meta.po.d sha256.po.d sha256-meta.po.d serpent.po.d serpent-met a.po.d twofish.po.d twofish-meta.po.d yarrow256.po.d yarrow_key_event.po.d sexp. po.d sexp-format.po.d sexp-transport.po.d sexp-transport-format.po.d bignum.po.d bignum-random.po.d sexp2bignum.po.d pkcs1.po.d pkcs1-rsa-md5.po.d pkcs1-rsa-sha 1.po.d rsa.po.d rsa-sign.po.d rsa-verify.po.d rsa-md5-sign.po.d rsa-md5-verify.p o.d rsa-sha1-sign.po.d rsa-sha1-verify.po.d rsa-encrypt.po.d rsa-decrypt.po.d rs a-keygen.po.d rsa-compat.po.d rsa2sexp.po.d sexp2rsa.po.d dsa.po.d dsa-sign.po.d dsa-verify.po.d dsa-keygen.po.d sexp2dsa.po.d pgp-encode.po.d rsa2openpgp.po.d buffer.po.d buffer-init.po.d realloc.po.d nettle-internal.po.d aesdata.po.d desd ata.po.d shadata.po.d make: fatal errors encountered -- cannot continue *** Error code 1
Stop in /usr/home/kiwi/Desktop/Downloads/Pike-v7.6-snapshot. kiwi@pc-system6:~/Desktop/Downloads/Pike-v7.6-snapshot$
It would also be helpful if people downloaded the latest generated build from http://pike.ida.liu.se/generated/pikefarm/packages/7.6/latest and reported any concerns.
Seems to compile well on FreeBSD 6.0-RC1 SMP... excepted the need for specifying threads library or the autoconf will blown itself and add -pthread and -lc_r thread library...
So with CONFIGUREARGS="--with-thread-library=-lpthread" this is working well
eg:
make => ok make doc => ok make snapshot => not ok (strange error with ansi thing as reported before.)
I didn't have time yet to try with Caudium, but this is on my todo...
/Xavier
-- Quand on essaye continuellement, on finit par y arriver. Donc, plus ca rate, plus on a de chance que ca marche... (Proverbe Shadok)
In the last episode (Oct 19), Xavier Beaudouin said:
Important note the nettle bundle don't work on a machine that doesn't have nettle installed :
make: fatal errors encountered -- cannot continue *** Error code 1
I was under the impression that Nettle's Makefile will only work with GNU make. That was the case a few versions ago, but I haven't tested with 1.13.
In the last episode (Oct 19), Xavier Beaudouin said:
Important note the nettle bundle don't work on a machine that doesn't have nettle installed :
make: fatal errors encountered -- cannot continue *** Error code 1
I was under the impression that Nettle's Makefile will only work with GNU make. That was the case a few versions ago, but I haven't tested with 1.13.
I have just tried to make it with gnu make.
I can confirm this is working.
Maybe you should add a note about the needs of using gnu make on systems that don't have gnu make as standart "make" utility...
/Xavier
In the last episode (Oct 19), Xavier Beaudouin said:
Important note the nettle bundle don't work on a machine that doesn't have nettle installed :
make: fatal errors encountered -- cannot continue *** Error code 1
I was under the impression that Nettle's Makefile will only work with GNU make. That was the case a few versions ago, but I haven't tested with 1.13.
That requirement should (AFAIK) be gone in Nettle 1.13.
Well, we have silently been working on a 7.6 release for the last few weeks, with e.g some fixes for the latest bundle versions yesterday. I've had reports on problems with generating the configure files from the CVS version, which is the only issue that currently blocks a release AFAIK. It would be nice if everyone took a fresh CVS-check out and made
make
Pike 7.6 from CVS today 11h55 CEST, darwin 8.4.0, gcc 4.0.0.
If you don't have nettle already installed (or not properly) on your system, make will fail while configuring Nettle. It seems it's not able to use Nettle as a bundle.
----8<----8<----8<----8<---- ################################################### ## Configuring module: Nettle ## Installation dir:
checking for C compiler default output file name... a.out checking whether the C compiler works... yes checking whether we are cross compiling... no checking for suffix of executables... checking for suffix of object files... (cached) o checking whether we are using the GNU C compiler... (cached) yes checking whether /Users/bertrand/Code/Pike/cvs/Pike/7.6.200510191155/ build/darwin-8.2.0-power-macintosh/smartlink gcc accepts -g... (cached) yes checking for /Users/bertrand/Code/Pike/cvs/Pike/7.6.200510191155/ build/darwin-8.2.0-power-macintosh/smartlink gcc option to accept ANSI C... (cached) none needed checking if we are using TCC (TenDRA C Compiler)... (cached) no checking how to run the C preprocessor... gcc -E checking for mpz_init in -lgmp... no checking for __mpz_init in -lgmp... no checking for __gmpz_init in -lgmp... yes checking for nettle_md5_init in -lnettle... no checking for md5_init in -lnettle... no Bundle nettle not available in /Users/bertrand/Code/Pike/cvs/Pike/ 7.6.200510191155/bundles. configure: error: Cannot compile --with-nettle without the Nettle library. It is highly recommended that you install a working Nettle library on your system as that will add features to Pike required by many applications. These features include ciphers, hash algorithms and random generators. If you know that you do not need cryptographic functions, you can re-run configure with the option --without-nettle instead. Nettle is available at http://www.lysator.liu.se/~nisse/ nettle/.
configure: error: /bin/bash -norc '/Users/bertrand/Code/Pike/cvs/Pike/ 7.6.200510191155/src/post_modules/Nettle/configure' failed for Nettle configure: error: /bin/bash -norc '/Users/bertrand/Code/Pike/cvs/Pike/ 7.6.200510191155/src/post_modules/configure' failed for post_modules make: *** [configure] Error 1 ---->8---->8---->8---->8----
With nettle properly installed on the system, it seems to be fine, while there are many warnings related to bundles:
----8<----8<----8<----8<---- /usr/bin/ld: warning -L: directory name (/Users/bertrand/Code/Pike/ cvs/Pike/7.6.200510191155/build/darwin-8.2.0-power-macintosh/bundles/ lib) does not exist ---->8---->8---->8---->8----
make doc make snapshots
Same as Jonas.
make verify:
Breaks here:
----8<----8<----8<----8<---- Part1: Character by character test Done. 98376 tests.
Part2: Canonical Order Test Done. 4224 tests. Subresult: 102741 tests, 0 failed, 0 skipped Doing tests in tlib/testsuite (24 tests) Subresult: 24 tests, 0 failed, 0 skipped Doing tests in tlib/7.0/modules/testsuite (2 tests) Subresult: 2 tests, 0 failed, 0 skipped Doing tests in tlib/7.2/modules/testsuite (10 tests) Subresult: 10 tests, 0 failed, 0 skipped Doing tests in tlib/7.4/modules/testsuite (1 tests) Subresult: 1 tests, 0 failed, 0 skipped Doing tests in tlib/7.4/modules/Crypto.pmod/testsuite (206 tests) Subresult: 206 tests, 0 failed, 6 skipped Doing tests in tlib/modules/testsuite (323 tests) Subresult: 323 tests, 0 failed, 1 skipped Doing tests in tlib/modules/ADT.pmod/testsuite (147 tests) Subresult: 147 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Calendar.pmod/testsuite (416 tests) Subresult: 416 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Crypto.pmod/testsuite (310 tests) Subresult: 310 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Filesystem.pmod/testsuite (3 tests) Subresult: 3 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Graphics.pmod/Graph.pmod/testsuite (6 tests) Subresult: 6 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Parser.pmod/LR.pmod/testsuite (1 tests) Subresult: 1 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Parser.pmod/XML.pmod/testsuite (15 tests) Subresult: 15 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Pike.pmod/testsuite (35 tests) Subresult: 35 tests, 0 failed, 35 skipped Doing tests in tlib/modules/Protocols.pmod/XMLRPC.pmod/testsuite (5 tests) Subresult: 5 tests, 0 failed, 0 skipped Doing tests in tlib/modules/SSL.pmod/testsuite (3 tests) Subresult: 3 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Standards.pmod/testsuite (84 tests) Subresult: 84 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Standards.pmod/ASN1.pmod/testsuite (18 tests) Subresult: 18 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Stdio.pmod/testsuite (24 tests) Subresult: 24 tests, 0 failed, 0 skipped Doing tests in tlib/modules/String.pmod/testsuite (118 tests) Subresult: 118 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Tools.pmod/testsuite (177 tests) Subresult: 177 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Web.pmod/testsuite (107 tests) Subresult: 107 tests, 0 failed, 0 skipped Doing tests in tlib/modules/Yabu.pmod/testsuite (37 tests) Subresult: 37 tests, 0 failed, 0 skipped Failed tests: 1. Total tests: 150381 (100 tests skipped) Finished tests at Wed Oct 19 17:28:40 2005 make[2]: *** [verify] Error 1 make[1]: *** [compile] Error 2 make: *** [verify] Error 2 ---->8---->8---->8---->8----
Hope this helps.
A checkout from CVS will not include the bundles, you have to download them separately.
On 19 oct. 05, at 17:50, Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) @ Pike (-) developers forum wrote:
A checkout from CVS will not include the bundles, you have to download them separately.
Doesn't the sources in the bundles directory count?
----8<----8<----8<----8<---- $ ls Pike/7.6/bundles/ CVS nettle-1.10.ptar.gz nettle-1.13.ptar.gz gmp.txt nettle-1.12.ptar.gz nettle.txt ---->8---->8---->8---->8----
Are the bundles supposed to work when running the xenoclient?
Those are not sources. Sources are however included in xenofarm packages, yes.
Runned xenoclient Nettle not being installed, and Nettle is functionnal as a bundle (darwin 8.2.0/gcc 4.0).
<url:http://pike.ida.liu.se/development/pikefarm/result.xml? id=355_225&pike=7_6>
Doesn't the sources in the bundles directory count?
----8<----8<----8<----8<---- $ ls Pike/7.6/bundles/ CVS nettle-1.10.ptar.gz nettle-1.13.ptar.gz gmp.txt nettle-1.12.ptar.gz nettle.txt ---->8---->8---->8---->8----
No. the ptars are patches applied to upstream before compiling the bundle.
Darwin 8.4? That would be 10.4.4 which I doubt anyone outside Apple has access to. :-)
Your "make verify" output only shows the ending message "Failed tests: 1" but not the place where it happens. Is it test_resolve.pike that's the culprit for you as well?
Darwin 8.4? That would be 10.4.4 which I doubt anyone outside Apple has access to. :-)
Typo, sorry :) I must have been thinking to gcc 4 while typing this.
Your "make verify" output only shows the ending message "Failed tests: 1" but not the place where it happens. Is it test_resolve.pike that's the culprit for you as well?
I should have included some more lines in my other mail, then:
----8<----8<----8<----8<---- Doing tests in post_modules/Unicode/testsuite (4 tests) test 1, line 1 Part0: Specific cases Done. 138 tests.
[here goes what i pasted in the other message] ---->8---->8---->8---->8----
The only one occurence of test_resolve.pike i have is there:
----8<----8<----8<----8<---- Begin tests at Wed Oct 19 21:58:16 2005 Doing tests in testsuite (10772 tests) test 500, line 2815 Testing vtables and scope access. 1:2 2:6 3:20 4:73 5:280 6:1118 7:4590 8:19279 /Users/bertrand/Code/Pike/cvs/Pike/7.6.200510191155/src/testsuite.in: 10260: Test 10425 (shift 2) failed to return proper results. 1: 2: array a() 3: { 4: Stdio.File p = Stdio.File(); 5: object pid = Process.create_process(RUNPIKE_ARRAY + 6: ({ "/Users/bertrand/Code/Pike/ cvs/Pike/7.6.200510191155/src/test_resolve.pike" }), 7: ([ "stdout":p->pipe (Stdio.PROP_IPC) ])); 8: __signal_watchdog(); 9: string results = lower_case(p->read()); 10: __signal_watchdog(); 11: int err = pid->wait(); 12: int ok; 13: int failed; 14: if (((sscanf(results, "%*sok:%d", ok) != 2) + 15: (sscanf(results, "%*sfailed:%d", failed) != 2)) == 2) { 16: // No "ok", and no "failed". 17: return ({ err, results }); 18: } 19: return ({ ok, failed }); 20: } 21:
o->a(): ({ /* 2 elements */ -1, "" })
.................................................. Testing end of heap ... Testing beginning of heap ... Waiting ................................ Removing call outs ......................... Subresult: 36958 tests, 1 failed, 7 skipped Doing tests in modules/_Charset/testsuite (1468 tests) ---->8---->8---->8---->8----
Yes, that's the same:
10260: Test 10425 (shift 2) failed to return proper results. [...]
Here you see the test number for the failing test as well as the source code and the actual output.
Running the test_resolve.pike script directly in gdb shows that it's the GL module that's broken. It's executing glGetIntegerv like this:
#0 0x92da3554 in glGetIntegerv () #1 0x020c9544 in f_glGet () #2 0x065dd96c in ?? () #3 0x0001761c in mega_apply (type=3377, args=-1073753040, arg1=0xbfffcf80, arg2=0x0) at /Users/jonasw/junk/pike/Pike-v7.6-snapshot/src/interpret.c:2006 #4 0x000952b4 in call_pike_initializers (o=0x20ac40, args=1684488) at /Users/jonasw/junk/pike/Pike-v7.6-snapshot/src/object.c:326
...and that symbol is taken from OpenGL.framework:
(gdb) info symbol 0x92da3530 glGetIntegerv in section LC_SEGMENT.__TEXT.__text of /System/Library/Frameworks/OpenGL.framework/Versions/A/Libraries/libGL.dylib
Now, the odd thing is that the .so file itself is linked to another GL library, namely the X11R6 dylib:
dark-castle:GL $ otool -L module.so module.so: /System/Library/Frameworks/Foundation.framework/Versions/C/Foundation (compatibility version 300.0.0, current version 567.12.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 88.2.1) /System/Library/Frameworks/CoreServices.framework/Versions/A/CoreServices (compatibility version 1.0.0, current version 18.0.0) /usr/X11R6/lib/libGL.1.dylib (compatibility version 1.2.0, current version 1.2.0) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ /usr/X11R6/lib/libXext.6.dylib (compatibility version 6.4.0, current version 6.4.0) /usr/X11R6/lib/libX11.6.dylib (compatibility version 6.2.0, current version 6.2.0)
Don't know if this mix is the cause of the problem but it does seem strange. Has anyone ever had the GL module working on OS X, and if so was that with or without X11 libraries installed at the same time?
Maybe a different module is linked against the OpenGL.framework (SDL?) and whichever gets loaded first "wins"?
pike-devel@lists.lysator.liu.se