Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #88701 > unrolled thread

Exception Handling

Back to article view | Back to comp.lang.python

Contents

  Exception Handling Palpandi <palpandi111@gmail.com> - 2015-04-09 02:31 -0700
    Re: Exception Handling Chris Angelico <rosuav@gmail.com> - 2015-04-09 20:36 +1000
    Re: Exception Handling Steven D'Aprano <steve+comp.lang.python@pearwood.info> - 2015-04-10 01:05 +1000
    Re: Exception Handling dieter <dieter@handshake.de> - 2015-04-10 08:11 +0200
    Re: Exception Handling Palpandi <palpandi111@gmail.com> - 2015-04-10 01:39 -0700
      Re: Exception Handling Chris Angelico <rosuav@gmail.com> - 2015-04-10 19:00 +1000
      Re: Exception Handling dieter <dieter@handshake.de> - 2015-04-11 07:39 +0200

#88701 — Exception Handling

Hi all,

Is there any way to roll back or undo changes which are all done before exception occurs.

[toc] | [next] | [standalone]

#88706

On Thu, Apr 9, 2015 at 7:31 PM, Palpandi <palpandi111@gmail.com> wrote:
> Hi all,
>
> Is there any way to roll back or undo changes which are all done before exception occurs.

In Python itself? Not directly; there are no facilities for undoing
Python code. But if you're serious about integrity, you probably want
to be (or already are) using a database, eg PostgreSQL, and storing
data in there. In that case, you can follow a fairly straight-forward
model, and one that some database connection modules even make easy
for you:

with dbconnection:
    # do stuff that might affect the database
    # and might throw an exception

As you leave the 'with' block, Python will tell the database
connection "hey, we got out without a problem", or "hey, we're leaving
here because of an exception". In the first case, the database will
commit, so everything you've done really happens. In the second, it'll
roll back, which will undo everything perfectly.

But this applies only to the database. For this to work, you have to
organize your code around that; make sure that everything that matters
is in the database, and local objects can't ever need to be "undone".

ChrisA

[toc] | [prev] | [next] | [standalone]

#88725

On Thu, 9 Apr 2015 07:31 pm, Palpandi wrote:

> Hi all,
> 
> Is there any way to roll back or undo changes which are all done before
> exception occurs.

Not automatically. You have to program it yourself.



-- 
Steven

[toc] | [prev] | [next] | [standalone]

#88770

Palpandi <palpandi111@gmail.com> writes:

> Is there any way to roll back or undo changes which are all done before exception occurs.

You are mostly interested in such a function when you handle
persistent data (data which persists across program activations -
otherwise, you can always quit the current program and restart it again).

Python usually does not handle persistent data by itself but
uses some kind of library or external system - e.g. a
relational database. Those external components may support
the concept of "transaction". Their primary function
is to ensure a consistent persistant state in view of concurrent
modifications. As a side effect, they support a "rollback"
which rolls back all modifications to persistent data made by the
rolled back transaction.

With some limitations, you can use the Python modules
"ZODB3" and "transaction" (available on "PyPI") to use
transactions in your program - even if you are not interested
in persistent data. Its "rollback" will not undo all changes but only
changes to "ZODB object"s.

[toc] | [prev] | [next] | [standalone]

#88771

Thanks for your responses. And I have one more question.

This is the situation. A python application which reads data from one .xml file and generates some other file. Here what kind of exceptions occur?

[toc] | [prev] | [next] | [standalone]

#88774

On Fri, Apr 10, 2015 at 6:39 PM, Palpandi <palpandi111@gmail.com> wrote:
> Thanks for your responses. And I have one more question.
>
> This is the situation. A python application which reads data from one .xml file and generates some other file. Here what kind of exceptions occur?

All sorts of exceptions. The best way to find out is to write your
code without any try/except clauses, and then run it. When an
exception happens, it'll be printed out on the console, and the
program terminated. Then you can decide how your program should cope
with that. Catch an exception *only* if you can actually deal with it
inside your code; otherwise, just let it bubble up.

ChrisA

[toc] | [prev] | [next] | [standalone]

#88803

Palpandi <palpandi111@gmail.com> writes:

> Thanks for your responses. And I have one more question.
> This is the situation. A python application which reads data from one .xml file and generates some other file. Here what kind of exceptions occur?

Ideally, you would not get any exceptions.

In the not ideal case, you may get exceptions from the XML parser
(in case, your XML input file is not valid), the "os" module
(for permission or filesystem related problems) and potentially
exceptions (such as "AttributeError") from your own program logic
(in case it would be faulty).

I support Chris' recommendation to not care about exceptions intially
(as they are exceptions - and do not occur frequently) and
decide what to do should they really occur.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web