Now it should always be possible to add a new record,
no matter how the table is defined or what the new data default is.
The default new data setting will be removed in the next commit.
Rewrite the import SQL code using a similar method to the one used in
executeQuery(). This makes the code much easier to read and removes the
last remnants of that C code stolen from some demo application.
Read and write the settings only from the preferences dialog.
Remove all the copies of some settings which were stored in nearly every
dialog class individually.
Simplify the settings dialog code by removing all those not really
needed slots.
Put the image editor into a scrollarea so the dialog doesn't screw up
when importing a large image file.
When importing a file add a filter to the file dialog to only show image
files.
By default, use the overwrite mode in the hex edit widget. This way text
and hex editor start in the same mode, making it less confusing to press
the INS key to change the mode.
When editing the data using the hex editor also update the text editor.
Allow the user to change between the text editor and the hex editor.
Allow the user to change between insert and overwrite mode in both
editors.
This makes it possible to enter any binary data into an empty cell.
Only show the old text edit widget when there is actually a text to be
edited. If the data is a image file show the image instead of some
random garbage. If it is some other binary data show the hex editor.
Store the data of a DB cell in a QByteArray, i.e. a binary data type,
instead of putting it in a QString, thus converting it to a UTF8 string.
Rewrite the reading and writing of DB cells to correctly handle binary
data containing 0x00 bytes.
Change the edit dialog to actually do all the data checks etc. on a
currently invisible QHexEdit widget instead of a QTextEdit.
All these changes combined make it possible to actually store binary
data without it being corrupted. You can for example import pictures
now, export them and actually open the exported file. So this is an
improvement.
There is also a "create table" ast walker which fills
info for the new sqlitetype objects.
A dependency to the antlr2 runtime was added.
The grammar most probably still contain bugs.
Why all this?
First writing grammars is fun and this is the only way
we can get all information for proper table editing + some time
in the future when the grammar is finished we can provide real
auto completion.
Fix the selection of entire columns in the browse tab table view. Even
when selecting multiple columns the data would be re-sorted instead of
selected; this is changed now.
