Path: csiph.com!v102.xanadu-bbs.net!xanadu-bbs.net!feeder.erje.net!eu.feeder.erje.net!newsfeed.xs4all.nl!newsfeed2.news.xs4all.nl!xs4all!post.news.xs4all.nl!not-for-mail
MIME-Version: 1.0
In-Reply-To: <CAPTjJmq=dydhFEooRHqG2UaHTREjJSyqk9McuDYhc8PRjSzTgA@mail.gmail.com>
References: <CANhUTb9tEnT=uTMhdA+Fd2zJTMauo1BWkX2dgcE0dG_H5arx9Q@mail.gmail.com> <CALvWhxtWyQw08TBVOTDhfFjm0=PP19QY3Yv-v74Q2T=ndyW91g@mail.gmail.com> <CAPTjJmq=dydhFEooRHqG2UaHTREjJSyqk9McuDYhc8PRjSzTgA@mail.gmail.com>
From: Chris Kaynor <ckaynor@zindagigames.com>
Date: Fri, 8 Aug 2014 16:55:39 -0700
Subject: Re: Newbie needing some help
To: Chris Angelico <rosuav@gmail.com>
Content-Type: multipart/alternative; boundary=047d7b5d912768d933050026f2df
Cc: "python-list@python.org" <python-list@python.org>
Precedence: list
Newsgroups: comp.lang.python
Message-ID: <mailman.12778.1407551506.18130.python-list@python.org>
Lines: 123
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:75923

--047d7b5d912768d933050026f2df
Content-Type: text/plain; charset=UTF-8

On Fri, Aug 8, 2014 at 4:38 PM, Chris Angelico <rosuav@gmail.com> wrote:

> On Sat, Aug 9, 2014 at 8:01 AM, Chris Kaynor <ckaynor@zindagigames.com>
> wrote:
> > However, I'd recommend re-raising the exception after rolling back the
> > transaction with a bare "raise" statement right after the db.rollback()
> - at
> > the absolute minimum, you should log the error. This will likely let you
> see
> > what your problem is.
>
> AIUI, aborting the process with an uncaught exception would roll back
> implicitly, so there's no need to "bare except, roll back, bare
> raise". But even if that's not the case, I would be looking at a
> try/finally for this, rather than a try/except/raise.
>

I would imagine that on connection close, the server would rollback any
uncommitted transactions, however that presumes that the connection close
is completed in a reasonably timely manner (with a try...finally for
closing, it should be).

This is one case where I think try...except/raise is better than
try...finally. If you use try...finally to rollback, you'd have to either
track whether you committed, or you'd rollback with no actions (or possibly
a committed action, which could have odd behavior, depending on the
implementation)

try:
    action
    commit
except:
    rollback
    raise

I feel is better than:

try:
    action
    commit
finally:
    rollback

or:

success = False
try:
    action
    commit
    success = True
finally:
    if success:
        rollback

and is very different than:

try:
    action
finally:
    commit

--047d7b5d912768d933050026f2df
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On F=
ri, Aug 8, 2014 at 4:38 PM, Chris Angelico <span dir=3D"ltr">&lt;<a href=3D=
"mailto:rosuav@gmail.com" target=3D"_blank">rosuav@gmail.com</a>&gt;</span>=
 wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D"">On Sat, Aug 9, 2014 at 8:01 =
AM, Chris Kaynor &lt;<a href=3D"mailto:ckaynor@zindagigames.com">ckaynor@zi=
ndagigames.com</a>&gt; wrote:<br>


&gt; However, I&#39;d recommend re-raising the exception after rolling back=
 the<br>
&gt; transaction with a bare &quot;raise&quot; statement right after the db=
.rollback() - at<br>
&gt; the absolute minimum, you should log the error. This will likely let y=
ou see<br>
&gt; what your problem is.<br>
<br>
</div>AIUI, aborting the process with an uncaught exception would roll back=
<br>
implicitly, so there&#39;s no need to &quot;bare except, roll back, bare<br=
>
raise&quot;. But even if that&#39;s not the case, I would be looking at a<b=
r>
try/finally for this, rather than a try/except/raise.<br></blockquote><div>=
<br></div><div>I would imagine that on connection close, the server would r=
ollback any uncommitted transactions, however that presumes that the connec=
tion close is completed in a reasonably timely manner (with a try...finally=
 for closing, it should be).</div>

<div><br></div><div>This is one case where I think try...except/raise is be=
tter than try...finally. If you use try...finally to rollback, you&#39;d ha=
ve to either track whether you committed, or you&#39;d rollback with no act=
ions (or possibly a committed action, which could have odd behavior, depend=
ing on the implementation)</div>

<div><br></div><div>try:</div><div>=C2=A0 =C2=A0 action</div><div>=C2=A0 =
=C2=A0 commit</div><div>except:</div><div>=C2=A0 =C2=A0 rollback</div><div>=
=C2=A0 =C2=A0 raise</div><div><br></div><div>I feel is better than:</div><d=
iv><br></div><div>try:</div><div>=C2=A0 =C2=A0 action</div>

<div>=C2=A0 =C2=A0 commit</div><div>finally:</div><div>=C2=A0 =C2=A0 rollba=
ck</div><div><br></div><div>or:</div><div><br></div><div>success =3D False<=
/div><div>try:</div><div>=C2=A0 =C2=A0 action</div><div>=C2=A0 =C2=A0 commi=
t</div><div>=C2=A0 =C2=A0 success =3D True</div><div>

finally:</div><div>=C2=A0 =C2=A0 if success:</div><div>=C2=A0 =C2=A0 =C2=A0=
 =C2=A0 rollback</div><div><br></div><div>and is very different than:</div>=
<div><br></div><div>try:</div><div>=C2=A0 =C2=A0 action</div><div>finally:<=
/div><div>=C2=A0 =C2=A0 commit</div></div></div>

</div>

--047d7b5d912768d933050026f2df--