|
Hi, thanks for valuable comments. The original description was meant as a kind of mathematical joke, since the task of proving a theorem is in general undecidable, so any theorem prover will get stuck forever on many (or most) inputs. I didn't realize at first that tis would actually discourage potentials users. Here is the corrected description: * Package name : eprover Version : 1.0.004 Upstream Author : Stephan Schulz <schulz@eprover.org> * URL : http://www.eprover.org/ * License : GPL-2 Programming Lang: C Description : Theorem prover for first-order logic with equality E is a fully automatic theorem prover for full first-order logic with equality. It accepts a mathematical specification and, optionally, a hypothesis, and tries to prove the hypothesis and/or find a saturation representing a (counter-)model for the specification. . E is based on a purely equational problem representation and implements a variant of the superposition calculus. Proof search can be guided with a multitude of options or a powerful automatic configuration mode. The system can process input in a number of different formats, including the standard TPTP-2 and TPTP-3 formats. It can generate proof objects in PCL2 or TPTP-3/TSTP format. . E is considered one of the most powerful and friendly automated theorem provers for first-order logic. It has consistently been among the top system in the major categories of the CASC system competition, and usually been the strongest free software system. Petr On Sunday 22 February 2009 16:18:58 Guus Sliepen wrote: > On Sun, Feb 22, 2009 at 09:57:04AM +0100, Petr Pudlak wrote: > > * Package name : eprover > > Description : The Equational Theorem Prover E > > That is not a description, that is just the name spelled out. A better > description would be "equational theorem prover". > > > E is an automated equational theorem prover. That means it is a program > > that you can stuff a mathematical specification (in first-order logic > > with equality) and a hypothesis into, > > So far so good... > > > and which will then run forever, using up all of your > > machines resources. Very occasionally it will find a proof for the > > hypothesis and tell you so ;-). > > If I read this, I think: "What kind of crap software is this that suffers > from infinite loops and only works for only a handful of equations? Why is > it in Debian if there are other theorem provers that are apparently > better?" > > > Release 1.0 is the culmination of a long development phase. Important > > changes vs. version 0.999 include the fixing of some bugs in definitional > > clausification for large problems and general cleanup. > > That information belongs in the upstream NEWS or Changes file, not in a > package description. > > > (Copied from the original documentation.) > > Ok. But this is not suited for a package description. Please create a > better description that is more serious, and explains what this prover can > and cannot do. Have a look at packages from similar software, like prover9, > if you need inspiration. |