iccifort-2020.4.304.eb fails to build CentOS 8.2 due to non-ascii character

[dithwick]

Building iccifort-2020.4.304.eb on CentOS 8.2 (Cascade Lake) using the system python 3.6 it failed with the following error message:

== configuring...
ERROR: Traceback (most recent call last):
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/main.py", line 117, in build_and_install_software
    (ec_res['success'], app_log, err) = build_and_install_one(ec, init_env)
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/framework/easyblock.py", line 3330, in build_and_install_one
    result = app.run_all_steps(run_test_cases=run_test_cases)
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/framework/easyblock.py", line 3229, in run_all_steps
    self.run_step(step_name, step_methods)
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/framework/easyblock.py", line 3084, in run_step
    step_method(self)()
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/easyblocks/generic/intelbase.py", line 296, in configure_step
    self.parse_components_list()
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/easyblocks/generic/intelbase.py", line 175, in parse_components_list
    mediaconfig = read_file(mediaconfigpath)
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/tools/filetools.py", line 195, in read_file
    txt = handle.read()
  File "/usr/lib64/python3.6/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe3 in position 306: ordinal not in range(128)

The build appears to be failing on the file

[build dir]/iccifort/2020.4.304/system-system/parallel_studio_xe_2020_update4_composer_edition/pset/mediaconfig.xml

which contains the line:

インテル(R) コンパイラー 19.1 Update 3 共通ファイル

The system on which this was failing on has the following locale settings:

$ locale
LANG=en_GB.UTF-8
LC_CTYPE="POSIX"
LC_NUMERIC="POSIX"
LC_TIME="POSIX"
LC_COLLATE="POSIX"
LC_MONETARY="POSIX"
LC_MESSAGES="POSIX"
LC_PAPER="POSIX"
LC_NAME="POSIX"
LC_ADDRESS="POSIX"
LC_TELEPHONE="POSIX"
LC_MEASUREMENT="POSIX"
LC_IDENTIFICATION="POSIX"
LC_ALL=POSIX

The build succeeds if run with:

LC_ALL= eb iccifort-2020.4.304.eb -Tr --rebuild

I did try changing intelbase.py

mediaconfig = read_file(mediaconfigpath)

to

mediaconfig = read_file(mediaconfigpath, mode='rb')

but I get the error message:

== configuring...
ERROR: Traceback (most recent call last):
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/main.py", line 117, in build_and_install_software
    (ec_res['success'], app_log, err) = build_and_install_one(ec, init_env)
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/framework/easyblock.py", line 3330, in build_and_install_one
    result = app.run_all_steps(run_test_cases=run_test_cases)
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/framework/easyblock.py", line 3229, in run_all_steps
    self.run_step(step_name, step_methods)
  File "/scrtp/avon/eb/software/EasyBuild/4.3.2/lib/python3.6/site-packages/easybuild/framework/easyblock.py", line 3084, in run_step
    step_method(self)()
  File "/tmp/eb-m4oscdka/included-easyblocks-ax0h0gw6/easybuild/easyblocks/generic/intelbase.py", line 296, in configure_step
    self.parse_components_list()
  File "/tmp/eb-m4oscdka/included-easyblocks-ax0h0gw6/easybuild/easyblocks/generic/intelbase.py", line 176, in parse_components_list
    available_components = re.findall("<Abbr>(?P<component>[^<]+)</Abbr>", mediaconfig, re.M)
  File "/usr/lib64/python3.6/re.py", line 222, in findall
    return _compile(pattern, flags).findall(string)
TypeError: cannot use a string pattern on a bytes-like object

Hope that helps.

[dithwick]

I get the same error with impi-2019.9.304-iccifort-2020.4.304.eb that is fixed with LC_ALL= at the front. I'm guessing all the intel modules will produce the same error.

[dithwick]

Just to flag up that unsurprisingly it affects other packages such as Python-2.7.18-GCCcore-10.2.0.eb

[boegel]

@dithwick Can you share the traceback you're getting with Python?

[dithwick]

@boegel Sorry, completely forgot to do this https://gist.github.com/dithwick/072e309b56b856e1642269b237570d13

[dithwick]

Works fine if LC_ALL= is added beforehand.