Re: [Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > On Thu, 16 Aug 2007, William Allen Simpson wrote: > You don't have a 2.0.* client. That's an unreleased development client. We have for a long time been running a fresh checkout of stable branches for pubserver games, since this way we could get the most important bug fixes out effective immediately. I do not think pubserver is working at the moment, though. - Per ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > On Thu, 16 Aug 2007, William Allen Simpson wrote: > If there were packet specific capabilities There are. See doc/README.delta - Per ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > Pepeto _ wrote: > You are wrong, the 2.0 branch is supposed be stable. All the user with a > 2.0.* client would be able to play on a 2.0.* server without > compatibility problems. > You don't have a 2.0.* client. That's an unreleased development client. Moreover, should you insist that all 2.0.x clients work forever with all 2.0.x servers, then you'll just have to live with the crashes the bug fixes prevent. I for one vow to never fix another 2.0 bug. After all, 2.0.9 crashes on the first turn! A fine thing to preserve ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > Marko Lindqvist wrote: > On 16/08/07, William Allen Simpson <[EMAIL PROTECTED]> wrote: >> There is no documented packet specific capability string. > > version.in, common/cap.c > That is not a documented packet specific capability string. It is neither: 1) documented nor 2) packet specific Traditionally, version.in is for releases, and should not be updated more than once per release cycle. It is not accessed except during configure. You are misusing long understood industry terminology. If there were packet specific capabilities, they would number in the hundreds or thousands. And would have been included in the packets.def file, to be generated for every handler, and every individual packet. For example, like tilespec capabilities, in each and every file. >> IT'S VERY DANGEROUS TO RUN DEVELOPMENT CODE AGAINST RELEASED SERVERS. > > The very idea of capability string is to make sure that incompatible > server/client do not accept connections from each other. Then every development version.in needs a capability that is *never* in the released version.in. Because nobody should willy-nilly be compiling development code and running it as or against a public server! It's really not my problem that this project cannot get releases out the door in a timely fashion, so poor saps in the field try to hack something together as their own pseudo-release. That's the fault of certain persons that disappear for months at a time, leaving the makefiles, configuration, and code in a sorry state. > IT'S VERY STUPID TO INSIST DOING SAME MISTAKE AFTER IT HAS BEEN > POINTED OUT TO YOU. > It's not a mistake. And you only mentioned such a thing in a very recent posting. The code of which he complains is from some time ago. ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > On 16/08/07, Pepeto _ <[EMAIL PROTECTED]> wrote: > > at revision 13068, 13069 and 13070, ATTRIBUTE_CHUNK_SIZE had been changed. > But, ATTRIBUTE_CHUNK_SIZE is used to define packets. Yes, this has to reverted from S2_1 and S2_0 immediately. I have no access to my development machine now. And trunk needs new manditory capability. - ML ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > On 16/08/07, William Allen Simpson <[EMAIL PROTECTED]> wrote: > > http://bugs.freeciv.org/Ticket/Display.html?id=39575 > > > Pepeto _ wrote: > > IT'S VERY DANGEROUS TO CHANGE PACKET DEFINITIONS WITHOUT REDEFINE > > CAPABILITY STRING. > > > There is no documented packet specific capability string. version.in, common/cap.c > IT'S VERY DANGEROUS TO RUN DEVELOPMENT CODE AGAINST RELEASED SERVERS. The very idea of capability string is to make sure that incompatible server/client do not accept connections from each other. ...and if we are supposed to yell: IT'S VERY STUPID TO INSIST DOING SAME MISTAKE AFTER IT HAS BEEN POINTED OUT TO YOU. - ML ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
[Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > > [wsimpson - Jeu. Aoû. 16 13:48:36 2007]: > > Pepeto _ wrote: > There is no released version with this new code. > > IT'S VERY DANGEROUS TO RUN DEVELOPMENT CODE AGAINST RELEASED SERVERS. > > You want to run development code, you also have to use servers from the > exact same commit. BY DEFINITION!!! You are wrong, the 2.0 branch is supposed be stable. All the user with a 2.0.* client would be able to play on a 2.0.* server without compatibility problems. ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > Pepeto _ wrote: > IT'S VERY DANGEROUS TO CHANGE PACKET DEFINITIONS WITHOUT REDEFINE > CAPABILITY STRING. > There is no documented packet specific capability string. There is no released version with this new code. IT'S VERY DANGEROUS TO RUN DEVELOPMENT CODE AGAINST RELEASED SERVERS. You want to run development code, you also have to use servers from the exact same commit. BY DEFINITION!!! > Please make something fast. Thank you. > That's entirely up to you. I tried to get folks to officially release an updated 2.0.10 (PR#39441), but no joy. ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
[Freeciv-Dev] (PR#39575) Redefinition of packets crash
http://bugs.freeciv.org/Ticket/Display.html?id=39575 > at revision 13068, 13069 and 13070, ATTRIBUTE_CHUNK_SIZE had been changed. But, ATTRIBUTE_CHUNK_SIZE is used to define packets. PACKET_PLAYER_ATTRIBUTE_CHUNK=47; pre-send, sc,cs,handle-via-packet UINT32 offset, total_length, chunk_length; /* to keep memory management simple don't allocate dynamic memory */ MEMORY data[ATTRIBUTE_CHUNK_SIZE:chunk_length]; end This cause nowadays lot of server's crash in metaserver. The half of the players have ATTRIBUTE_CHUNK_SIZE = 1400, the other ATTRIBUTE_CHUNK_SIZE = 2*1024. IT'S VERY DANGEROUS TO CHANGE PACKET DEFINITIONS WITHOUT REDEFINE CAPABILITY STRING. Please make something fast. Thank you. ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
