Complete.Org: Mailing Lists: Archives: freeciv-dev: January 2002:
[Freeciv-Dev] Re: CVS's civserver core dumps
Home

[Freeciv-Dev] Re: CVS's civserver core dumps

[Top] [All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
To: Vasco Alexandre Da Silva Costa <vasc@xxxxxxxxxxxxxx>
Cc: Ben Webb <ben@xxxxxxxxxxxxxxxxxxxxxx>, Paul Zastoupil <paulz@xxxxxxxxxxxx>, <freeciv-dev@xxxxxxxxxxx>
Subject: [Freeciv-Dev] Re: CVS's civserver core dumps
From: "Pieter J. Kersten" <kersten@xxxxxxxxxx>
Date: Mon, 14 Jan 2002 10:22:19 +0100 (CET)

On Sat, 12 Jan 2002, Vasco Alexandre Da Silva Costa wrote:

> On Fri, 11 Jan 2002, Ben Webb wrote:
>
> > On Fri, 11 Jan 2002, Pieter J. Kersten wrote:
> >
> > > Whenever you try to build a city in the latest CVS version, the server
> > > core dumps. See the following backtrace.
> >
> >     By "whenever" I take it you mean under all circumstances. I can't
> > reproduce this bug with brief testing with latest CVS (GTK+ isometric or
> > Xaw client, server with standard options, one human player). Could you be
> > more specific?
> >
> >     Both this crash and the client crash reported by Paul seem to be
> > caused by the generalised improvements code that Vasco committed yesterday
> > and today. However, it is very different from my original code, so it's
> > entirely possible that the changes have introduced bugs. Can you reproduce
> > these crashes with the original impr-gen patches "effect-init-v2.patch"
> > and "effect-implement-v2.patch" against CVS of Jan 8th? (They obviously
> > won't apply against current CVS.) The patches are available at
> > http://freecivac.sourceforge.net/ or alternatively as PR#1140 and PR#1142.
> > If so, then I will investigate further.
>
> The crash Paul found was in improvement_redundant(). IIRC i have not
> changed that function. There is however a chance some of the other things
> i commited introduced that bug in some obscure way.
>
> In particular, i didn't add the changes to update_all_effects() and
> is_effect_activated() in effect-implement-v2.patch.
>
> Since most of the patches consisted of fairly insulated code that merely
> constructed and updated some lists without it being used anywhere yet, i
> assumed they wouldn't impact things elsewhere. Unless there is a buffer
> overrun somewhere.
>
> My limited testing didn't show up any bugs. Could you please provide a
> savegame that reproduces the bug?
>

It seems that the latest CVS commits solved this particular problem. I can
build cities again.

Thanx,
-- 
Pieter J. Kersten



[Prev in Thread] Current Thread [Next in Thread]