Re: [pysqlite] Seg Fault Only When Table Contains Data

On Monday 02 June 2008 6:00 pm, Rich Shepard wrote:
>    I could use some independent ideas to help me solve this problem. It
> follows on this past weekend's thread on writing a variable number of
> fields to a SQLite table.
>
>    Having solved that problem, I loaded the initial columns in the wxPython
> grid widget and saved them to the table. (The save method was presented in
> my last message on that thread.) But, this morning when I invoked the
> application after adding more functionality, it bombed with a segmentation
> fault as soon as I selected the project data file to open. Running strace
> (and gdb on the core dump) gave me no insight into the source of the
> problem. Howevever, by commenting out code in various modules and trying to
> run the application within winpdb I can identify where the application
> crashes, but not why it does.
>
>    There is a module, dbMethods.py, that selects records from the SQLite
> tables and transfers them to lists for quick access by various other
> modules in the application. One section of the OpenDB() method is:
>
>      # existing conditions and alternatives data
>      self.appData.cur.execute("SELECT * from Data")
>      self.appData.altData = self.appData.cur.fetchall()
>
> If I comment out either (or both) of these two lines, the application opens
> and loads normally; no seg fault. But, I can get an error about a
> zero-length list when only the SELECT statement is missing.
>
>    I also use the Publisher() methods to pass messages among modules. This
> method has the line:
>
>      Publisher().sendMessage(self.appData.projOpen, data=None)
>
> If I comment that out, no seg fault; with it in, the app crashes.
>
>    The dataPage.py module (with the grid widget) has an analogous
> directive:
>
>      Publisher().subscribe(self.loadData, self.appData.projOpen)
>
> Again, if I comment out that line there's no seg fault.
>
>    The data in the SQLite table look OK to me; e.g.,
>
> sqlite> select * from Data limit 10;
> Aesthetics||AmbientLight||||||||||||||||||||||||||||||
> Aesthetics||Viewsheds||||||||||||||||||||||||||||||
> AirQuality|FugitiveDust|Levels||||||||||||||||||||||||||||||
> AirQuality|FugitiveDust|Persistence||||||||||||||||||||||||||||||
> AirQuality|FugitiveDust|SpecialConsiderations||||||||||||||||||||||||||||||
> AirQuality|FugitiveDust|Timing||||||||||||||||||||||||||||||
> AirQuality|Particulates|Amount||||||||||||||||||||||||||||||
> AirQuality|Particulates|HealthEffects||||||||||||||||||||||||||||||
> AirQuality|RegionalHaze|DegreeAffected||||||||||||||||||||||||||||||
> CulturalHeritage||Quantity||||||||||||||||||||||||||||||
>
>    If I 'DELETE from Data' so the table's empty, the application runs. I
> can then initialize that table and save the results. But, then it seg
> faults when I try to run it again.
>
>    I'm completely stumped here; have no idea what's going on or why. Any
> thoughts you might have are more than welcome!
>
> Rich

Something about the data is causing the problem, either a data type mismatch 
or too much data. Might try:
self.appData.cur.execute("SELECT * from Data limit 10")
to see if its a excess data problem.

Try the following to see if its an empty string problem:
for c in range(33):
         if self.dataGrid.GetCellValue(r,c) == None:
           row.append('None)

-- 
Adrian Klaver
[email protected]