Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-04-02 Thread James Peach
On Mar 25, 2014, at 11:48 PM, Nick Kew wrote: > On Tue, 2014-03-25 at 20:25 -0600, Phil Sorber wrote: > An API change like that affects existing plugins and could leave us needing some ugly #ifdef crap to support both before- and after- TS versions. Can I make a plea to avoid tha

Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-26 Thread Phil Sorber
On Wed, Mar 26, 2014 at 12:48 AM, Nick Kew wrote: > On Tue, 2014-03-25 at 20:25 -0600, Phil Sorber wrote: > > > > > An API change like that affects existing plugins and could leave us > > > > needing some ugly #ifdef crap to support both before- and after- TS > > > > versions. Can I make a plea

Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-25 Thread Nick Kew
On Tue, 2014-03-25 at 20:25 -0600, Phil Sorber wrote: > > > An API change like that affects existing plugins and could leave us > > > needing some ugly #ifdef crap to support both before- and after- TS > > > versions. Can I make a plea to avoid that: maybe a new function name, > > > and migrate t

Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-25 Thread Phil Sorber
On Tue, Mar 25, 2014 at 8:39 AM, Leif Hedstrom wrote: > > On Mar 25, 2014, at 12:16 AM, Nick Kew wrote: > > > On Mon, 2014-03-24 at 20:39 -0600, Leif Hedstrom wrote: > > > >> with a consistent transfer of ownership of both "body" and > "content_type", both will be free()'d when the core no longe

Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-25 Thread Leif Hedstrom
On Mar 25, 2014, at 12:16 AM, Nick Kew wrote: > On Mon, 2014-03-24 at 20:39 -0600, Leif Hedstrom wrote: > >> with a consistent transfer of ownership of both “body” and “content_type”, >> both will be free()’d when the core no longer needs them. > > ISTR that being a minor "gotcha". There's a

Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-24 Thread Nick Kew
On Mon, 2014-03-24 at 20:39 -0600, Leif Hedstrom wrote: > with a consistent transfer of ownership of both “body” and “content_type”, > both will be free()’d when the core no longer needs them. ISTR that being a minor "gotcha". There's a comment in the Ironbee code about the body being freed so

Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-24 Thread Leif Hedstrom
On Mar 24, 2014, at 3:26 PM, James Peach wrote: > On Mar 24, 2014, at 9:45 AM, Leif Hedstrom wrote: > >> 3. In order to make the TSHttpTxnErrorBodySet() API easier to use, I’ve >> changed the content type argument to not transfer ownership. The common >> (well, exclusive) pattern using this

Re: [DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-24 Thread James Peach
On Mar 24, 2014, at 9:45 AM, Leif Hedstrom wrote: > Hi all, > > > We currently have two APIs to set an “error” body. One was added at Yahoo > many years ago, and it was done for the wrong reasons (premature optimization > being one of them). I’ve made a patch that unifies these two APIs into

[DISCUSS} Changes to TSHttpTxnErrorBodySet and TSHttpTxnSetHttpRetBody

2014-03-24 Thread Leif Hedstrom
Hi all, We currently have two APIs to set an “error” body. One was added at Yahoo many years ago, and it was done for the wrong reasons (premature optimization being one of them). I’ve made a patch that unifies these two APIs into one, such that you can use TSHttpTxnErrorBodySet() anywhere you