note. SRC_URI="http://${PN}.googlecode.com/files/${P}.tar.gz mirror://pypi/${PN:0:1}/${PN}/${P}.tar.gz" archtester html5lib # ebuild html5lib-0.95.ebuild test merge >>> /usr/share/doc/html5lib-0.95/README.bz2 * Compilation and optimization of Python modules for CPython 2.7 ... [ ok ] >>> dev-python/html5lib-0.95 merged. good to go
With the below patch, this fails rather a lot of tests for me. --- html5lib-0.90.ebuild 2011-03-06 16:59:02.000000000 +0100 +++ html5lib-0.95.ebuild 2012-03-01 17:27:46.000000000 +0100 @@ -7,20 +7,20 @@ PYTHON_USE_WITH="xml" SUPPORT_PYTHON_ABIS="1" RESTRICT_PYTHON_ABIS="3.*" +DISTUTILS_SRC_TEST="nosetests" inherit distutils DESCRIPTION="HTML parser based on the HTML5 specification" HOMEPAGE="http://code.google.com/p/html5lib/ http://pypi.python.org/pypi/html5lib" -SRC_URI="http://${PN}.googlecode.com/files/${P}.zip" +SRC_URI="mirror://pypi/${PN:0:1}/${PN}/${P}.tar.gz" LICENSE="MIT" SLOT="0" KEYWORDS="amd64 x86 ~x86-fbsd" IUSE="examples" -DEPEND="app-arch/unzip - dev-python/setuptools" +DEPEND="dev-python/setuptools" RDEPEND="" src_install() { I think some of this might have to do with chardet and/or lxml being installed.
I tried it again; the test suite I can't get to work gentoo64 html5lib-0.95-2.7 # python2.6 html5lib/tests/runtests.py ---------------------------------------------------------------------- Ran 0 tests in 0.000s OK Use of nosetests pulls 85 errors consistently. RESTRICT="test"
(In reply to comment #1) > I think some of this might have to do with chardet and/or lxml being > installed. Indeed. With both chardet and lxml installed, I get 7988 errors, 35 failures. Without chardet, I get 0 errors, 85 failures. Without chardet and lxml 2 errors, 12 failures. It seems like a very poor codebase or test suite.
Created attachment 311027 [details, diff] ebuild patch http://code.google.com/p/html5lib/issues/detail?id=201&colspec=ID%20Type%20Status%20Priority%20Milestone%20Owner%20Summary%20Port The reading of the issues makes for sad viewing. -- The issues are reported in reverse chronological order. -- The only issues reviewed and dealt with are those submitted by the project members themselves. -- The last example of just that was May 1st. making reference to "we don't handle the case where we get given a file-like object that returns Unicode strings." which is consistent with this mass failure rate. Perhaps it's reasonable in this case to do as I have in this patch This version has been out for months.
The bug you filed upstream is very hard to follow. Some tips: - Try not to list Gentoo-specific commands in upstream bug reports. - Explain the symptoms concisely, in complete English sentences.
(In reply to comment #3) > It seems like a very poor codebase or test suite. Forget I said this please. The 7988 errors are what caused me to say that, but that appear to be caused by a warning that chardet is producing: UnicodeWarning: Unicode equal comparison failed to convert both arguments to Unicode - interpreting them as being unequal
I just bumped this with the tests disabled. Thanks for trying.