processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • PyTables only supports fixed-width string columns in tables. The analysis toolkit, Release 0.13.1 – numexpr 2.2.2 fixes incompatiblity in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • PyTables only supports fixed-width string columns in tables. The thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatiblity in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • Once a table is created columns (DataFrame) are fixed; only exactly

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. 23.8. HDF5 (PyTables) 715 pandas: powerful Python data analysis thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatibility in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • PyTables only supports fixed-width string columns in tables. The thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatibility in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. 824 Chapter 24. IO Tools (Text, CSV, HDF5, ...) pandas: powerful thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatibility in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • Once a table is created its items (Panel) / columns (DataFrame) thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatibility in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • Once a table is created its items (Panel) / columns (DataFrame) thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatibility in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • Once a table is created its items (Panel) / columns (DataFrame) thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatibility in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported

processes, you may want to use fsync() before releasing write locks. For convenience you can use store.flush(fsync=True) to do this for you. • Once a table is created its items (Panel) / columns (DataFrame) thanks @tavistmorph and @numpand – numexpr 2.2.2 fixes incompatibility in PyTables 2.4 (GH4908) – flush now accepts an fsync parameter, which defaults to False (GH5364) – unicode indices not supported