From: Steve O. <st...@so...> - 2010-12-08 15:00:27
|
----- "Dean Michael Berris" <mik...@gm...> wrote: > Hi Steve, > > On Wed, Dec 8, 2010 at 2:01 PM, Steve Obbayi <st...@so...> > wrote: > > Hi Dean, I am now back, still jet lagged :) > > Great, I hope you had a good vacation. :) > > > Anyway as I was re-familiarizing myself with the code I noticed that > if the user has no > > network and tries to use the http client they get an exception and > the http client quits. > > > > This is by design -- all operations with the synchronous client (like > get, post, put, head, delete_) all throw in case something goes wrong > at any stage of the process (except when reading, an eof causes the > read from the socket to terminate under normal circumstances). > > With the asynchronous client, when you try to use any of the wrappers, > and there was something wrong in the process of getting the response, > it will throw the error that is associated with the error_code > received. > > > I found this out when looking into how to implement the retry on > connection failure feature. > > > > At this point I have not gone further to root out the cause because > a question arose in the process. Should the http client be implemented > with all client connections being made under a try/catch block? Or > should the http client handle these kind of situations internally? Or > is this a bug in the http client. > > > > The client is designed to throw when operations are performed, this is > so that the user of the client can do what they deem necessary in > situations where the client encounters an exceptional condition -- > network errors are considered exceptional conditions. ;) > > > I am using version 0.8 with visual C++ 2010. > > > > You will get this no matter what compiler you use. ;) > > Also, if you are going to work on anything in cpp-netlib, please work > against the latest development branch. Patches submitted against 0.8 > will pretty much be useless with the amount of changes that have > happened between 0.8 and 0.9-devel. :D > > HTH > > -- > Dean Michael Berris > deanberris.com > Okay thanks, for confirming this to me. -- Steve Obbayi SKYPE: sobbayi http://sobbayi.com http://blog.sobbayi.com |