Debugging Gramps

From Gramps
Revision as of 00:03, 1 December 2023 by Bamaustin (talk | contribs) (Undo revision 100692 by Bamaustin (talk))
Jump to: navigation, search
Gramps-notes.png
This document is targeted at developers who are using a development environment with Gramps.

See the developers portal about setting up a Gramps development environment.

Tango-Dialog-information.png
Please note when doing debugging,

that this document mentions src/gramps.py and Gramps.py. Since you never have both in a single Gramps setup, please use src/gramps.py when running Gramps 3.x and earlier, and Gramps.py for 4.x and newer.


An overview on how to debug Gramps.

Hard crash

When a hard crash occurs, you typically have no idea where the crash occurs. It is important to know the line where the crash occurs. So if you are able to reproduce the crash, restart Gramps with the command:

python -m trace -t Gramps.py

This will usually generate vast terminal output and slow the system down for that, so redirecting output to a file is usually good:

python -m trace -t Gramps.py >/tmp/trace.out

Then check the file "/tmp/trace.out" and save if needed, or redirect to somewhere else in the earlier step.

Add debug statements

Gramps is run with the optimize flag.

python -O Gramps.py

This gives you the option of adding debug statements. You can use the __debug__ variable or the assert statement for this. This allows us to add code to Gramps that will be printed out when Gramps runs without the optimize flag:

python Gramps.py

More info: [1]

Use the log infrastructure

Gramps has built in the python log infrastructure. Gramps runs with logging level logging.DEBUG to stderrh.

More info: Logging system

Short: in your code add data to the logger with: log.warning(), log.error(), log.info()...

Start Gramps with the --debug flag:

python Gramps.py --debug="name_of_the_logger"

This is useful when working in different parts, adding info output, and selecting on the command line with --debug the logger you want to see output with.

Use profiling

Gramps has a convenience hook to allow you to do profiling. Profiling gives an overview how many times methods are called in a code fragment and how long each one took. It helps to find performance bottlenecks.

An example usage:

Add at the top of the file you want to use profiling:

from gramps.gen.utils.debug import profile

Then, suppose you want to profile a save function on one of the editors. The save is called due to a connect done in the method _connect_signals. So, change the connect to save to a new method testsave you make:

#self.define_ok_button(self.top.get_widget("ok"), self.save)
self.define_ok_button(self.top.get_widget("ok"), self.testsave)

Now add the testsave function as follows, where you use the profile convenience function:

   def testsave(self, *obj):
       profile(self.save, *obj)

Then run Gramps, every time you save, the profiler kicks in and prints to command line a report with the time for each function.

So in short: replace the call to a function/method to calling profile with as first parameter the function/method, and other parameters the parameters of the function/method.

That's it.

See also GEPS_016:_Enhancing_Gramps_Processing_Speed for a sample of Profile Analysis.

Use the winpdb python debugger

Gramps-notes.png
Note

There are some issues with the winpdb debugger. For a workaround see bug ticket: 2564

Winpdb (Ubuntu: sudo apt-get install winpdb) is a graphical interface for the python debugger. Start it with:

winpdb Gramps.py
Winpdb with Gramps loaded

Now, in File menu, select 'Open Source', and open the source file you want to debug. Now all debug options are possible. Eg, go to a line in the file with the cursor, and click the run to line button. The debugger will run to that point, and left show you all defined local variables with their value, as well as the stack frame.

Try it!


Use gdb C debugger

Gramps-notes.png
Note

Knowledge of the C programming language required.

With GObject introspection, much more can go wrong on a lower level, causing C errors. For these gdb can be used. You should install some debug libraries like libglib2.0-0-dbg, python-gobject-dbg, ... .

Then, you start in a terminal gdb in the directory where Gramps.py is stored with

gdb python

You should see something like:

 GNU gdb (Ubuntu/Linaro 7.2-1ubuntu11) 7.2
 Copyright (C) 2010 Free Software Foundation, Inc.
 License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
 This is free software: you are free to change and redistribute it.
 There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
 and "show warranty" for details.
 This GDB was configured as "x86_64-linux-gnu".
 For bug reporting instructions, please see:
 <http://www.gnu.org/software/gdb/bugs/>...
 Reading symbols from /usr/bin/python...Reading symbols from /usr/lib/debug/usr/bin/python2.7...done.
 done.
 (gdb) 

You just type now:

run Gramps.py

and then navigate gramps to the segmentation fault. After that, you see something like

Program received signal SIGSEGV, Segmentation fault.
0x00007ffff5e7226b in g_type_check_value () from /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
(gdb)

Now you request the backtrace, to see the code that leads to the error:

(gdb) bt

which will print the backtrace. You are now ready to use your C knowledge to fix the bug!

Trace GObject/GTK warnings

GTK has several possible warning levels, which might pop up on the terminal, without visible problems in Gramps. To trace these, do the following in the :

gdb python

and at the gdb prompt indicate to stop on log output:

b g_log if log level < 32
r Gramps.py

Now Gramps will stop on serious messages, and you can use

bt

to get a backtrace of the C stack showing where the message originated and what called the function, etc.

You can force a hard crash on warnings by setting G_DEBUG.

For example:

 LC_ALL=C G_DEBUG=fatal-warnings python -m trace -t Gramps.py

Different values possible: [2]

Learn More

To learn more about debugging python and C debugging with gdb, see [3] and [4].

Learning C tutorials

The Python language implementation is written in the C programming language.