You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Windows: mkconcore mis-reads and then crashes on non-ASCII node names (text files opened without an encoding) #706
mkconcore.py opens its text files (the GraphML, the generated build/run/debug scripts, .iport/.oport, Dockerfile templates) without an encoding, so on Windows they use the system code page (cp1252). A workflow with non-ASCII node names such as ctrl_é.py or плант.py then breaks in two ways:
The UTF-8 GraphML is read as cp1252, so names turn into mojibake ("плант.py") and the build logs "Script ... not found in sourcedir".
Once the names are read correctly, writing the generated scripts fails with UnicodeEncodeError ('charmap' codec can't encode characters) at fbuild.write('mkdir '+containername).
Same root cause as #666 (BOM read as ""). I reproduced both by running mkconcore with a cp1252 default for open(). Proposal: read the GraphML as utf-8-sig and use encoding="utf-8" for every text file mkconcore opens. ASCII-only output is unchanged. I have a fix and a test ready.
Activity