[BioPython] 1.21 installation difficulty

[Leighton Pritchard]

Thanks, Michiel and Brad,

That worked brilliantly. I had to get the CVS version of 1.21 due to the 
absent Martel, but under Python2.2 the mingw32 and a test Borland install 
both went pretty well (one failure and two errors in the test suite...

Error: test_GenBank

Traceback (most recent call last):
   File "test_GenBank.py", line 121, in ?
     cur_seqrecord = gb_dict[key]
   File 
"e:\Applications\Python2_2\Lib\site-packages\Bio\GenBank\__init__.py", line 
88, in __getitem__
     return self._parser.parse(File.StringHandle(data))
   File 
"e:\Applications\Python2_2\Lib\site-packages\Bio\GenBank\__init__.py", line 
229, in parse
     self._scanner.feed(handle, self._consumer)
   File 
"e:\Applications\Python2_2\Lib\site-packages\Bio\GenBank\__init__.py", line 
1244, in feed
     self._parser.parseFile(handle)
   File "E:\APPLIC~1\PYTHON~1\lib\site-packages\Martel\Parser.py", line 
328, in parseFile
     self.parseString(fileobj.read())
   File "E:\APPLIC~1\PYTHON~1\lib\site-packages\Martel\Parser.py", line 
356, in parseString
     self._err_handler.fatalError(result)
   File ".\xml\sax\handler.py", line 38, in fatalError
Martel.Parser.ParserPositionException: error parsing at or beyond character 0

[The test_GenBank error only seems to come up in the CVS installation, by 
the way. The test build with the Borland compiler from the published source 
.zip file didn't throw the this error.]

Failure: test_SVDSuperimposer - AssertionError, line 285 of run_tests.py, 
in compare_output

Output: '[[5.16518888e+001 ....
Expected: '[[5.16518888e+01 ....

Could this be a typo in the assertion? There are zeros missing in the 
exponents of the other expected array values, too...

Error: test_copen

Traceback (most recent call last):
   File "test_copen.py", line 14, in ?
     handle = copen.copen_fn(print_args, *(range(2) + ['a', 'b', 'c']))
   File "E:\APPLIC~1\PYTHON~1\lib\site-packages\Bio\MultiProc\copen.py", 
line 67, in copen_fn
     pid = os.fork()
AttributeError: 'module' object has no attribute 'fork')

I found the same problems as Michiel with Python2.3c - the Borland compiler 
gets about as far as the MS compiler with the egenix source:

building 'mx.TextTools.mxTextTools.mxTextTools' extension
mx\TextTools\mxTextTools\mxTextTools.c:
Error E2209 E:\APPLIC~1\PYTHON~1\include\pyconfig.h 37: Unable to open 
include file 'io.h'
Error E2209 E:\APPLIC~1\PYTHON~1\include\pyconfig.h 163: Unable to open 
include file 'stdio.h'
Error E2209 E:\APPLIC~1\PYTHON~1\include\Python.h 27: Unable to open 
include file 'limits.h'
Error E2209 E:\APPLIC~1\PYTHON~1\include\Python.h 42: Unable to open 
include file 'stdio.h'
Fatal F1003 E:\APPLIC~1\PYTHON~1\include\Python.h 44: Error directive: 
"Python.h requires that stdio.h define NULL."
*** 5 errors in Compile ***
creating build\temp.win32-2.2
creating build\temp.win32-2.2\Release
creating build\temp.win32-2.2\Release\mx
creating build\temp.win32-2.2\Release\mx\TextTools
creating build\temp.win32-2.2\Release\mx\TextTools\mxTextTools
creating build\temp.win32-2.2\Release\mx\TextTools\mxTextTools\mxTextTools
e:\Applications\bcc55\bin\bcc32.exe -c /tWM /O2 /q /g0 -UHAVE_STRPTIME 
-UHAVE_TIMEGM -Imx/TextTools/mxTextTools 
-IE:\APPLIC~1\PYTHON~1\include 
-obuild\temp.win32-2.2\Release\mx\TextTools\mxTextTools\mxTextTools\mxtexttools.obj 
mx\TextTools\mxTextTools\mxTextTools.c
error: command 'bcc32.exe' failed with exit status 1

Attempting to build a Windows distribution from the mingw32-compiled code 
throws up the same problems as Michiel encountered, which I got round the 
same way. When I attempted to install the package on my machine (over the 
existing installation), I also came across some file permission problems 
with writing the EUtils/DTD files, which I corrected by changing read-only 
permissions (are these default on XP or something?) to read-write on the 
Bio sub-tree.

Many thanks for all your help,