I agree this should be fixed... and your solution looks workable. Not sure if Robert would rather a linked list, a char *[] array, or the resizing buffer... Regardless we should be using the rm_eo-rm_so to determine length and probably can use realloc instead of malloc.
The cookie generation patch was posted and should work for you, but has not been merged into mainline. Joe > -----Original Message----- > From: Francisco Ruiz [mailto:[email protected]] > Sent: Tuesday, November 15, 2011 4:42 PM > To: [email protected] > Subject: [Pound Mailing List] AddHeader directive honored > > Hi, > > I send a patch to the list so that it can be considered and merged to > the main code if accepted. > > We need to add several headers to the request but it seems that > AddHeader just honors the last occurrence of the directive. Even more, > several AddHeader directives doesn't rise an error or warning. All of > them are processed and the string created with previus strdup() call > seems not to be freed. > > The patch sent: > * modify config.c to concatenate all the headers in the LISTENER > struct member add_head with the needed "\r\n". > * modify the comment to LISTENER struct member add_head to reflect > that AddHeader works also with HTTP an not only HTTPS, in pound.h > * modify man page pound.8 to reflect the changes. > > I hope it can be of any help for others. > > On the other hand, I would like to know about the patch that someone > send to the list to get sticky sessions by injecting cookies in the > replies to client, as other commercial products do. Are there any plans > to merge it? > > Thanks for pound and to all the community arround it. > > -- > Francisco Ruiz > [email protected] > > Ahorrar papel protege el medio ambiente -- To unsubscribe send an email with subject unsubscribe to [email protected]. Please contact [email protected] for questions.
