Home Original page

[Python-Dev] ConfigParser mangles keys with special chars

Ethan Furman ethan at stoneleaf.us
Fri Apr 25 21:12:13 CEST 2014 
On 04/25/2014 11:54 AM, Fred Drake wrote:
> On Fri, Apr 25, 2014 at 2:45 PM, Terry Reedy <tjreedy at udel.edu> wrote:
>> I leave it to someone to carefully read the doc, but a brief glance
>> indicates "There are nearly as many INI format variants as there are
>> applications using it. configparser goes a long way to provide support for
>> the largest sensible set of INI styles available." So I wonder whether the
>> thought-to-be-buggy behavior is actually buggy with respect to *all* the
>> supported styles or just some of them.
>
> Given that most variants are completely undocumented, answering this
> is sufficiently intractable for me to call it intractable.
>
> We've also run into compatibility issues because some applications
> treated the "parser" object as an in-memory configuration database
> throughout the process lifetime, updating values with non-string
> values of various sorts.  Given the age of the module, the existing
> number of uses of undocumented accidents of implementation is very
> large.  Fixing "bugs" like this has an excellent track record of
> uncovering these uses, so... we've grown a bit wary.  It's not unheard
> of for projects to fork their own copies of configparser because of
> changes to the stdlib version.  (No, I haven't done a census of such
> cases, but I do know of at least one in a popular package.)

Seems reasonable.

Perhaps an enhancement request then:  a way to provide a regex that keys must match, with an exception raised when a key 
doesn't.  That way the safety belt could be used when desired.

--
~Ethan~
More information about the Python-Dev mailing list