Opening FASTA file fails

[Tom Goddard]

Opening the attached fasta file works the first time. Then when I quit ChimeraX and open it again it fails with the following error. From then on it always fails to open the file.

UCSF ChimeraX version: 0.1 (2017-01-06)
OpenGL version: 4.1 NVIDIA-10.14.20 355.10.05.15f03
OpenGL renderer: NVIDIA GeForce GT 650M OpenGL Engine
OpenGL vendor: NVIDIA Corporation

open ~/ucsf/data/ihm-2016/nup84/extra_data/alignments/nup84_3ikoC_3jroC_3f3fG.fa

Traceback (most recent call last):
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/cmd_line/tool.py", line 158, in execute
cmd.run(cmd_text)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/core/commands/cli.py", line 2255, in run
result = ci.function(session, kw_args)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/core/commands/open.py", line 108, in open
models = session.models.open(paths, format=format, name=name, kw)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/core/models.py", line 351, in open
session, filenames, format=format, name=name, kw)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/core/io.py", line 368, in open_multiple_data
models, status = open_data(session, fspec, format=format, name=name, kw)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/core/io.py", line 328, in open_data
models, status = open_func(session, stream, dname, kw)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/core/toolshed/init.py", line 1872, in open_cb
return open_shim(*args, kw)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/core/toolshed/init.py", line 1869, in open_shim
return f(*args, format_name=format_name, kw)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/seqalign/init.py", line 41, in open_file
return open_file(*args, kw)
File "/Users/goddard/Desktop/ChimeraX Jan 6 2017.app/Contents/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/chimerax/seqalign/parse.py", line 35, in open_file
raise ValueError("No sequences found in %s file '%s'!" % (format_name, fname))
ValueError: No sequences found in FASTA file 'nup84_3ikoC_3jroC_3f3fG.fa'!

[Tom Goddard]

Tested with January 6 2017 nightly build on Mac OS 10.12.2

[Tom Goddard]

If I do a make install in bundles/alignments in my development build then it will open the FASTA file correctly the first time I run ChimeraX, and then fails to open it every subsequent time I run ChimeraX.

If I do the make install, run ChimeraX and quit without opening the FASTA file, then run ChimeraX a second time, that also fails to open the FASTA. This suggests the problem is something to do with Toolshed caching something when it first fines a new bundle.

[Eric Pettersen]

The first time the alignment file is opened, the repr() of the stream passed to my open/parse function is:

<_io.TextIOWrapper name='/Users/pett/src/chimera/libs/MultAlignViewer/test/nonassoc.fa' mode='rt' encoding='utf-8'>

The second time it's opened, what's passed in is:

<_io.BufferedReader name='/Users/pett/src/chimera/libs/MultAlignViewer/test/nonassoc.fa'>

[Conrad Huang]

Fixed in 10697fb.

The data format encoding information was not saved in the toolshed cache, so restarting ChimeraX causes sequence files to be opened in binary mode instead of text.