Re: Weird problems with C extension and bytea as input type - Mailing list pgsql-general

From Merlin Moncure
Subject Re: Weird problems with C extension and bytea as input type
Date
Msg-id AANLkTinyQihT0BAeuoMdL+cO1zxOFZmvNDyiLFErpmGR@mail.gmail.com
Whole thread Raw
In response to Re: Weird problems with C extension and bytea as input type  (dennis jenkins <dennis.jenkins.75@gmail.com>)
Responses Re: Weird problems with C extension and bytea as input type
Re: Weird problems with C extension and bytea as input type
List pgsql-general
On Wed, Mar 23, 2011 at 9:04 AM, dennis jenkins
<dennis.jenkins.75@gmail.com> wrote:
> On Wed, Mar 23, 2011 at 5:08 AM, Adrian Schreyer <ams214@cam.ac.uk> wrote:
>>
>> you are right, it returns a char *.
>>
>> The prototype:
>>
>> char *function(bytea *b);
>>
>> The actual C++ function looks roughly like this
>>
>> extern "C"
>> char *function(bytea *b)
>> {
>>   string ism;
>>   [...]
>>   return ism.c_str();
>> }
>>
>
>
> Don't do that.  You are returning a pointer to an unallocated buffer
> (previously held by a local variable).  c_str() is just a const
> pointer to a buffer held inside "ism".  When ism goes out of scope,
> that buffer if freed.
>
> Either return "std::string", or strdup() the string and have the
> caller free that.  (but use the postgresql alloc pool function to
> handle the strdup.  I don't recall that function's name off the top of
> my head).

that would be pstrdup, and it's the way to go (you don't have to
pfree).  who says C doesn't have garbage collection?

merlin

pgsql-general by date:

Previous
From: Merlin Moncure
Date:
Subject: Re: PostgreSQL documentation specifies 2-element array for float8_accum but 3-element array expected
Next
From: Tom Lane
Date:
Subject: Re: Weird problems with C extension and bytea as input type