Version: 2.0.0_devel (using unspecified) Compiler: gcc (SUSE Linux) 4.4.1 OS: Linux To reproduce: - generate two ASCII test files with a specific, non-standard header configuration (lines to skip, line containing field names) - try to load the first one from the datawizard and check that the listed field names are not read correctly - come back to the first screen, and click the Configure... button. Set the appropriate settings, check the "Apply these settings as default" box and click OK - click next: the list of fields found should now be right - create one or two plots from that file to check that it reads the data correctly - launch the datawizard again and select the second file with the same non-standard formatting - click Next: the field list is wrong, the previously set default settings have not been obeyed
SVN commit 1117430 by brisset: When reading a file with the ASCII datasource, always load ASCII defaults before attempting to load file-specific settings (asciisourceconfig.cpp). Also avoid overriding those defaults with hard-coded values when no key exists for the given file (namedparameter.h). That issue is actually more a problem that the test in asciisourceconfig.cpp / AsciiSourceConfig::readGroup is not specific enough: fileName.isEmpty() triggers too easily => if the file has no settings stored, we still try to read its settings and since we find nothing we revert to hard-coded default values. With this patch, the constructor sets hard-coded defaults, then we override them with the ASCII source's general defaults, then with file-specific values, but only if there are some in QSettings. If you think this patch is wrong, we have to find a way to test for the existence of settings for the given file instead of just testing whether the filename is empty. BUG: 234396 M +2 -3 asciisourceconfig.cpp M +1 -3 namedparameter.h WebSVN link: http://websvn.kde.org/?view=rev&revision=1117430
Change version from 2.0.0_devel to 2.0.0 to simplify version numbering.
These bugs are solved with 2.0.0