I've had to drop the ~sparc keyword from hpricot-0.8 because it adds a dependency over dev-util/ragel that is not keyworded. Have fun! Diego
~sparc added for ragel. However, hpricot-0.8 has some problems: 1) Source file seems to be bad --- 'Failed on RMD160 verification', 'e89cf084628359d82a56fcaa180b1a44d7887219', 'c96bc6b560b45ce43790d3383aa41aa41b31729e') !!! Fetched file: hpricot-0.8.tgz VERIFY FAILED! !!! Reason: Failed on RMD160 verification !!! Got: e89cf084628359d82a56fcaa180b1a44d7887219 !!! Expected: c96bc6b560b45ce43790d3383aa41aa41b31729e Refetching... File renamed to '/usr/portage/distfiles/hpricot-0.8.tgz._checksum_failure_.LrzdKW' 2) If I force the ebuild to match this source file, tests fail: =========================== 1) Failure: test_xml_casing(TestAlter) [./test/test_alter.rb:64]: <"<root><wildCat>text</wildCat></wildCat><beanPole>gravity</beanPole></beanPole></root>"> expected but was <"<root><wildCat>text</wildCat><beanPole>gravity</beanPole></root>">. 2) Failure: test_scan_boingboing(TestParser) [./test/test_parser.rb:125]: <116> expected but was <110>. 72 tests, 6159 assertions, 2 failures, 0 errors rake aborted! Command failed with status (1): [/usr/bin/ruby18 -I"/var/tmp/portage/dev-ru...] (See full trace by running task with --trace) * * ERROR: dev-ruby/hpricot-0.8 failed. * Call stack: * ebuild.sh, line 49: Called src_test * environment, line 2464: Called die * The specific snippet of code: * $ruby $(type -p rake) test || die "testsuite failed"; * The die message: * testsuite failed ================================ Are these problems, to be expected, or a problem with ragel? Please advise.
Something is very wrong with the server that's providing these distfiles: >>> Downloading 'http://github.com/why/hpricot/tarball/0.8' --2009-05-08 18:29:19-- http://github.com/why/hpricot/tarball/0.8 Resolving github.com... 65.74.177.129 Connecting to github.com|65.74.177.129|:80... connected. HTTP request sent, awaiting response... 302 Moved Temporarily Location: http://waitdownload.github.com/why-hpricot-b35bc08d1e85f5c32471462fc25927b2379f45f6.tar.gz [following] --2009-05-08 18:29:19-- http://waitdownload.github.com/why-hpricot-b35bc08d1e85f5c32471462fc25927b2379f45f6.tar.gz Resolving waitdownload.github.com... 65.74.177.146 Connecting to waitdownload.github.com|65.74.177.146|:80... connected. HTTP request sent, awaiting response... 302 Moved Temporarily Location: http://download.github.com/why-hpricot-b35bc08d1e85f5c32471462fc25927b2379f45f6.tar.gz [following] --2009-05-08 18:29:19-- http://download.github.com/why-hpricot-b35bc08d1e85f5c32471462fc25927b2379f45f6.tar.gz Resolving download.github.com... 65.74.177.162 Connecting to download.github.com|65.74.177.162|:80... connected. HTTP request sent, awaiting response... 200 OK Length: 256331 (250K) [application/x-gzip] Saving to: `/home/gentoo/distfiles/hpricot-0.8.tgz' 100%[================================================================================================================================================================================================================================>] 256,331 439K/s in 0.6s 2009-05-08 18:29:20 (439 KB/s) - `/home/gentoo/distfiles/hpricot-0.8.tgz' saved [256331/256331] ('Failed on RMD160 verification', 'f8265ef39a496351f5e51fcc0300bbb64331e48f', 'c96bc6b560b45ce43790d3383aa41aa41b31729e') !!! Fetched file: hpricot-0.8.tgz VERIFY FAILED! !!! Reason: Failed on RMD160 verification !!! Got: f8265ef39a496351f5e51fcc0300bbb64331e48f !!! Expected: c96bc6b560b45ce43790d3383aa41aa41b31729e Refetching... File renamed to '/home/gentoo/distfiles/hpricot-0.8.tgz._checksum_failure_.62W07Z' Notice that it does NOT match fmccor's checksums either. There is a single word at offset 0x4 that differs between them: # diff -Nuar <(xxd </home/gentoo/distfiles/hpricot-0.8.tgz._checksum_failure_.62W07Z) <(xxd /home/gentoo/distfiles/hpricot-0.8.tgz._checksum_failure_.TZ31Nr) --- /dev/fd/63 2009-05-08 18:34:17.586867981 -0700 +++ /dev/fd/62 2009-05-08 18:34:17.586867981 -0700 @@ -1,4 +1,4 @@ -0000000: 1f8b 0800 abda 044a 0003 ec5b 6b73 db46 .......J...[ks.F +0000000: 1f8b 0800 447f 044a 0003 ec5b 6b73 db46 ....D..J...[ks.F 0000010: 96cd 67fc 8a36 bd55 7a14 0d5a 9265 c52a ..g..6.Uz..Z.e.* 0000020: cb33 b44c c7f4 c892 56a4 c749 b952 aa06 .3.L....V..I.R.. 0000030: d024 1181 0082 8768 e6f1 dff7 9cdb 0d90 .$.....h........ Whatever they are using to generate the tarballs is not making them consistently.
I start to feel like I hate github to my guts.
I've replaced the github fetching with a tarball I built myself, should be better now.
I noticed something interesting. I in the period of half an hour downloaded 3 copies all of the same file, all with different checksums. Upon un-gzipping them, the checksums are the same again. So GitHub probably have a bug in on-the-fly gzipping. (not tar itself)
http://support.github.com/discussions/repos/755-inconsistent-md5s-of-downloaded-files >> git archive --format=tar --prefix=#{user}-#{repo}-#{sha}/ --remote=#{path} #{sha} | gzip Methinks they're doing it on-the-fly instead of one-time and caching the result. for i in $( seq 0 1000 ); do git archive --format=tar master | gzip | md5sum; done You'll find that the md5sum printed varies exactly in time with your system clock.
>> We've added the -n option to gzip and the checksums are now consistent. Github returns to be usable :)
This now builds as expected on sparc. But on sparc, we see two test failures, thus: ================================================= Loaded suite /usr/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake/rake_test_loader Started ........F..........................................F.................... Finished in 6.772235 seconds. 1) Failure: test_xml_casing(TestAlter) [./test/test_alter.rb:64]: <"<root><wildCat>text</wildCat></wildCat><beanPole>gravity</beanPole></beanPole></root>"> expected but was <"<root><wildCat>text</wildCat><beanPole>gravity</beanPole></root>">. 2) Failure: test_scan_boingboing(TestParser) [./test/test_parser.rb:125]: <116> expected but was <110>. 72 tests, 6159 assertions, 2 failures, 0 errors ====================================== Please advise if this is a problem or not. I can't compare with hpricot-0.6 because it has no test suite, but I do not see the failures on amd64. (Both are running dev-lang/ruby-1.8.7_p160.)
I've just added hpricot-0.8.3, could you test again on sparc to see if the test failures are now gone?
This has been keyworded already...
