From news-rocq.inria.fr!univ-lyon1.fr!pasteur.fr!oleane!newsxfer3.itd.umich.edu!news.maxwell.syr.edu!news.new-york.net!news.stormking.com!bezzi@nemo.it Fri Jan 9 10:50:17 1998 Article: 8139 of rec.games.corewar Path: news-rocq.inria.fr!univ-lyon1.fr!pasteur.fr!oleane!newsxfer3.itd.umich.edu!news.maxwell.syr.edu!news.new-york.net!news.stormking.com!bezzi@nemo.it From: Beppe Bezzi Newsgroups: rec.games.corewar Subject: Corewarrior 64 Date: 8 Jan 1998 17:37:22 -0500 Organization: Storm King Ind. Inc. Lines: 758 Sender: server@news.stormking.com Distribution: world Message-ID: Reply-To: bezzi@nemo.it NNTP-Posting-Host: valhalla.stormking.com Originator: corewar-l@stormking.com X-Mailer: Windows Eudora Light Version 1.5.2 .xX$$x. .x$$$$$$$x. d$$$$$$$$$$$ ,$$$$$$$P' `P' , . $$$$$$P' ' .d b $$$$$P b ,$$x ,$$x ,$$x ,$$b $$. Y$$$$' `$. $$$$$$. $$$$$$ $$P~d$. d$$$b d d$$$ `$$$$ ,$$ $$$$$$$b $$$P `$ $$$b.$$b `Y$$$d$d$$$' . . a . a a .aa . a `$$$ ,$$$,$$' `$$$ $$$' ' $$P$XX$' `$$$$$$$$$ .dP' `$'$ `$'$ , $''$ `$'$ `Y$b ,d$$$P `$b,d$P' `$$. `$$. , `$$P $$$' Y $. $ $ $ Y..P $ `$$$$$$$' $$$P' `$$b `$$$P `P `$' `Y'k. $. $. $. $$' $. Issue 64 8 January 1998 ______________________________________________________________________________ Core Warrior is a newsletter promoting the game of corewar. Emphasis is placed on the most active hills--currently the '94 draft hill and the beginner hill. Coverage will follow where ever the action is. If you have no clue what I'm talking about then check out these five-star internet locals for more information: FAQs are available by anonymous FTP from rtfm.mit.edu as pub/usenet/news.answers/games/corewar-faq.Z FTP site is: ftp.csua.berkeley.edu /pub/corewar ftp.inria.fr/INRIA/Projects/para/doligez/cw/mirror. Web pages are at: (Please note new Stormking's address) http://www.koth.org/ ;Stormking http://www.ecst.csuchico.edu/~pizza/koth ;Pizza http://pauillac.inria.fr/~doligez/corewar/ ;Planar Newbies should check the stormking page for the FAQ, language specification, guides, and tutorials. Post questions to rec.games.corewar. All new players are infinitely welcome! If ftp.csua.berkeley.edu is unreachable, you can download pMARS at: Terry's web page--http://www.ncs.infi.net/~wtnewton/corewar/ Planar ftp site--ftp://ftp.inria.fr/INRIA/Projects/para/doligez/cw/mirror Fechter ftp site--ftp://members.aol.com/ofechner/corewar A collection of Bezzi's hints in the first issues is available at: ftp://ftp.volftp.vol.it/pub/pc/msdos/games/solutions/bbhints.zip Beppe Bezzi web page - http://www.aspide.it/freeweb/Bezzi ______________________________________________________________________________ Greetings. I'm a bit late with this issue, sorry, but i hope you'll be satisfied with the code of Nomolos and obvious and forgive me. I had some problems of space with this issue and i had to cut the beginners hill and the old hall of fame; my mailer strongly refused to paste other text. Best wishes of merry Christmas and Happy New Year to you all. --Beppe Bezzi _____________________________________________________________________________ Current Status of the Internet Pizza Server ICWS '94 Draft Hill: Hill Specs: coresize: 8000 max. processes: 8000 duration: after 80,000 cycles, a tie is declared. max. entry length: 100 minimum distance: 100 rounds fought: 200 instruction set: ICWS '94 Draft # %W / %L / %T Name Author Score Age 1 46.6/ 36.1/ 17.3 Alladins Cave P.Kline 157.0 6 2 41.1/ 29.1/ 29.9 Nomolos Ian Oversby 153.0 10 3 38.9/ 28.1/ 33.0 Fixed Ken Espiritu 149.8 5 4 36.6/ 24.4/ 39.0 Newt Ian Oversby 148.9 99 5 37.9/ 29.0/ 33.0 Vain Ian Oversby 146.8 21 6 36.8/ 27.0/ 36.2 Vigor Ken Espiritu 146.7 9 7 42.2/ 38.3/ 19.4 Electric Head Anton Marsden 146.1 108 8 45.6/ 45.8/ 8.6 He Scans Again P.Kline 145.5 1 9 42.1/ 40.2/ 17.7 Red Baron Christian Schmidt 144.0 64 10 33.7/ 23.4/ 43.0 Benchmark [94] Zul Nadzri 143.9 21 11 39.6/ 36.6/ 23.8 PAN-TAU-RA Christian Schmidt 142.6 6 12 36.5/ 30.5/ 33.0 obvious to those who know Robert Macrae 142.4 57 13 41.3/ 42.4/ 16.3 The Boys are Back in Town Philip Kendall 140.3 3 14 33.7/ 27.5/ 38.8 Pulp v0.5 Ian Oversby 139.9 56 15 34.8/ 30.2/ 35.1 Vengeance Robert Hale 139.3 56 16 38.4/ 38.0/ 23.6 Stasis David Moore 138.8 2 17 40.5/ 42.6/ 16.9 Silver Talon 1.2 Edgar 138.4 7 18 37.1/ 36.8/ 26.1 Tornado 4 Beppe Bezzi 137.4 23 19 28.2/ 19.4/ 52.4 Return Of The Jedimp John K W 137.1 353 20 42.5/ 48.2/ 9.4 Memories Beppe Bezzi 136.7 27 21 38.8/ 41.5/ 19.7 HnK II Ivan Krajnovic 136.2 5 22 34.1/ 32.7/ 33.1 Head or Tail Christian Schmidt 135.5 141 23 31.9/ 28.4/ 39.7 Ultraviolet-B Ken Espiritu 135.5 31 24 38.9/ 42.7/ 18.4 It's not a trick Christian Schmidt 135.2 12 25 33.9/ 33.0/ 33.1 Nine Seven Six M R Bremer 134.8 217 Age since last issue: 26 ( 31 last issue, 27 the issue before ) New warriors: 15 Turnover/age rate 60% Average age: 53 ( 63 last issue, 63 the issue before ) Average score: 142 ( 136 last issue, 136 the issue before ) The top 25 warriors are represented by 14 independent authors: Schmidt and Oversby with 4; Espiritu with 3; Kline and Bezzi with 2. All others with one warrior each. Paul Kline put his Alladins Cave on the top and here it remained all time, even if he tried very hard to do better in 94 test; let's see who will be the first to find the magic word ('open sesame' is for Ali Baba cave :-) ______________________________________________________________________________ 94 - What's New # %W / %L / %T Name Author Score Age 1 41.6/ 36.9/ 21.5 Alladins Cave P.Kline 146.4 2 2 36.0/ 28.1/ 36.0 Fixed Ken Espiritu 143.9 1 4 33.6/ 26.9/ 39.5 Vigor Ken Espiritu 140.2 5 5 34.4/ 29.7/ 35.8 Nomolos Ian Oversby 139.2 6 6 33.9/ 29.1/ 37.0 Vain Ian Oversby 138.8 17 9 37.8/ 41.4/ 20.8 Silver Talon 1.2 Edgar 134.1 3 10 36.6/ 39.4/ 24.0 It's not a trick Christian Schmidt 133.9 8 13 28.2/ 24.8/ 47.0 Benchmark [94] Zul Nadzri 131.5 17 17 39.6/ 49.6/ 10.8 Memories Beppe Bezzi 129.6 23 18 28.4/ 27.5/ 44.1 Ultraviolet-B Ken Espiritu 129.3 27 22 34.8/ 42.4/ 22.8 HnK II Ivan Krajnovic 127.1 1 24 31.8/ 37.2/ 31.0 Tornado 4 Beppe Bezzi 126.4 19 6 40.4/ 41.9/ 17.7 The Boys are Back in Town Philip Kendall 138.8 1 17 35.6/ 39.6/ 24.9 Stasis David Moore 131.6 1 9 43.6/ 47.5/ 9.0 He Scans Again P.Kline 139.6 1 More than half hill is changed. ______________________________________________________________________________ 94 - What's No More 26 35.6/ 43.5/ 20.9 Damage Incorporated Anton Marsden 127.7 373 26 1.0/ 0.0/ 3.0 Dim Wit Ian Oversby 6.1 35 26 39.1/ 50.4/ 10.5 He Scans Again P.Kline 127.8 76 26 32.1/ 42.1/ 25.8 PaperBoy Throws Some Rock Robert Hale 122.1 29 26 35.9/ 48.4/ 15.7 C^3 Christian Schmidt 123.5 51 26 1.7/ 1.6/ 0.7 Nomolos Ian Oversby 5.7 17 26 37.2/ 49.9/ 12.9 Conan the Barbarian Christian Schmidt 124.4 15 26 26.7/ 29.2/ 44.1 Merciless-A Ken Espiritu 124.2 56 26 32.7/ 43.1/ 24.1 Digitalis 2 Christian Schmidt 122.4 37 And something else that i missed :-( Damage incorporated is the leaves the hill and loses the * in the Hall of Fame ______________________________________________________________________________ 94 - What's Old # %W / %L / %T Name Author Score Age 20 23.5/ 19.0/ 57.5 Return Of The Jedimp John K W 128.1 353 19 30.6/ 32.4/ 37.0 Nine Seven Six M R Bremer 128.8 217 12 31.8/ 31.7/ 36.5 Head or Tail Christian Schmidt 131.9 141 14 35.6/ 40.3/ 24.1 Electric Head Anton Marsden 131.0 108 Return of the Jedimp is the oldest warrior now, Elrctric Head the new entry. ______________________________________________________________________________ No space left for Old Hall of Fame, no changes from last issue. ______________________________________________________________________________ NEW HALL OF FAME * means the warrior is still active. Pos Name Author Age Strategy 1 Probe Anton Marsden 403 Q^2 -> Bomber 2 Blur 2 Anton Marsden 396 Scanner 3 Damage Incorporated Anton Marsden 373 Q^2 -> Bomber 4 Return Of The Jedimp John K W 353 * Q^2 -> Stone/imp 5 unrequited love kafka 346 Q^2 -> Paper 6 Impish v0.2 Ian Oversby 345 Stone/imp 7 Gigolo Core Warrior staff 332 Q^2 -> Stone/imp 8 Falcon v0.3 Ian Oversby 275 P-warrior 9 Rosebud Beppe 218 Stone/imp 10 Nine Seven Six M R Bremer 217 * Q^2 -> Stone/imp 11 Q^2 Miro Anders Ivner 214 Q^2 -> Scanner/bomber 12 Instant Wolf 3.4 Edgar 205 P-warrior 13 Goldfinch P.Kline 201 P-warrior 14 Simple v0.4b Ian Oversby 197 QScan -> Stone/imp 15 Trident^2 John K W 195 Q^2 -> Stone/imp 16 ompega Steven Morrell 189 Stone/imp 17 Frogz Franz 172 Q^2 -> Paper 18 The Machine Anton Marsden 164 Scanner 19 Memories Beppe 152 Scanner 20 Head or Tail Christian Schmidt 141 * Q^2 -> Paper 21 Tiberius 3.1 Franz 130 Q^2 -> Paper 22 Solomon v0.8 Ian Oversby 116 Stone and scanner 23 CC Paper 3.3 Franz 107 Q^2 -> Paper 24 mrb-test m r bremer 106 ? 25 T.N.T. pro Maurizio Vittuari 105 Bomber Jedimps gains 3 spots, nine seven six 5, Head or tail 3; no new entries. ______________________________________________________________________________ Current Status of the Internet Pizza Server Beginner's Hill: Sorry no space for beginners hill :-( ______________________________________________________________________________ Extra Extra (I) Nomolos by Ian Oversby I wouldn't normally publish anything as flimsy as a p-spacer. However, maybe Kline does have a point that code-sharing will encourage others to jump in so I thought I would share Nomolos. I am interested to know how making the code public will effect it. > By the way, I think the hill is a bit weird at the moment > with many complex p-switchers. Testing results are very > different from real ones and changing boot distances a bit > often changes the score significantly. > > Then those damn tests and not scoring if the dead warrior > is less than 110 or something. I hate another warrior being > the same age as one of mine. Is it just my imagination or > did people stop playing when tests were introduced? Anyway, > that is quite enough from this soapbox... The original concept was quite simple. As the hill is so full of paper, kill the papers and then switch to a imp/paper to draw against everything else. I had quite an aggressive imp/paper that scored reasonably against most warriors, only suffering from HSA and Memories type things. I tried various warriors for the paper-killer including a Blur-style warrior, mini-HSA and CMP scanners but nothing did quite so well as one-shot scanners. I wrote a fairly flawed one a while back called Tsunami which also handled some imp/stones well. The result did OK, landing in the upper part of the hill but was easily beaten by Electric Head and HSA/Scorch. I then wanted to handle scanners with something else and change the paper to something a little more aggressive. Carbonite is very effective against HSA/Scorch so I chose that. The last point was the choice of paper. I was only really expecting paper/imps imp/stones and P-Spacers at this point so it had to perform quite well alround against scanners. The original Pulp is a little fragile against scanners so after a little experimentation I came up with a replacement that is OK against E-Head and isn't destroyed by scanners. For the p-space algorithm resetting on tie for the stone seems to work against pure scanners the best. I also thought it would be best to stay at the paper, even on losing. If I am brainwashed to this point by a scanner, maybe I will be brainwashed back to the stone again. It seems to work OK anyway. ;---------- That was the original plan. Then came several imp/papers against which, my one-shot is helpless. I wanted to add mini-HSA which also handles imp/papers but there wasn't enough space so I developed a new p-space brain and added mini-HSA. This did OK for a few challenges until people updated their imp/papers but the real blow was when the next Bloodhound appeared. This trashed Nomolos 25/140 so I made a minor tweak (sorry Christian :-) and now it takes hill-top again. I suspect not for very long though... ;redcode-94 ;name Nomolos ;author Ian Oversby ;strategy Stone, One-shot Scanner ;strategy then try to tie ;strategy v2 Added mini-HSA ;strategy One scanner is never enough :-) ;assert 1 ;-------- ; Paper ;-------- pdist equ 4000 ; Not really dist0 equ 5689 dist1 equ 1021 dist2 equ 3607 range1 equ 4723 range2 equ 6892 pboot spl.b 1, <-300 for 6 mov.i -1 silk1 spl.b @0, -1 mov.i bomba, }range1 mov.i bomba, >range1 mov.i {silk1, ptr,>ptr jmn.f kill,>ptr a add #mst+1,@mptr scan jmz.f a,-1, >1 ;----------------- ; One-shot scanner ;----------------- tgate equ (tloop-8) half equ 10 step equ (2*half) dstr equ 3001 tdist equ 700 tpos equ (cptr+1) tboot mov.i -303 dat.f #-25, #cptr+4-tgate spl.b #-20, #20 b1 spl.b #120, #120-half tloop sub.f tstep, b1 sne.i *b1, @b1 djn.f tloop, <-dstr tstep spl.b #-step, <-step mov.i @cptr, >tgate mov.i @cptr, >tgate cptr djn.b -2, {b1 end pcode ______________________________________________________________________________ Extra Extra (II) obvious to those who know By Robert Mcrae "obvious" -- Or, How Not to Name Your Warrior. obvious has just made it into middle age at 50, a good occasion to publish it. Besides, I've been working on a oneshot for *ages* and with just a few more papers on the hill... ;-) obvious is a Q^2 scan followed by an aggressive paper. The Q^2 is actually a new engine, which I wrote for convenience rather than power as it doesn't work any better than the Probe masterpiece. However, new constants make it slightly more programmer-friendly because there are now no accesses outside the table of pointers. This makes it relatively easy to modify or re-order the code. I've kept the Probe attack as it has a particularly neat bombing pattern. After lengthy analysis I came up with a full Tornado-style bomber which shaved a couple of cycles off the bomb run but it never justified the extra length and boredom quickly set in. If you want a longer scan, there are eight lines which are available but are currently commented out. I found that extra length gave away more losses to other Q^2s than it gained in wins; as far as I can see a longer run would only make sense with a less aggressive main warrior, or when there are fewer Q^2s on the hill. The paper is also a close copy of existing code, in this case CCPaper: len EQU 9 fcp EQU 3039 scp EQU 2365 tcp EQU 777 boot spl 1, <-3000 ; 9 processes together mov -1, 0 mov -1, 0 mov -1, 0 frog spl @0, -1 spl @0, -1 spl @0, -1 mov 2, <-1 jmp -1, <-10 dat <2667, <2667*2 The idea of CCPaper is to make an aggressive replicator. Each Silk carries a simple suicidal coreclear and as the battle progresses, more and more processes accumulate in these clears. This gives very effective coverage of core, and also means that an enemy warrior which manages to hop aboard the Silk replicator when it is over-written is not yet safe. Unless it manages to subvert the mechanism very effectively and spread widely within core, it may still be wiped by the other DAT clears. If it only establishes itself in a single clear then it will eventually wipe itself. Sad 1.0, my variant, makes two small modifications to this warrior. The first is to cut the number of processes from 9 to 7 by adding a second MOV instruction to the first stage replicator. This reduction in processes leads to a substantial increase in replication rates, as less time is wasted by the subsequent stages copying trailing DATs. I have found that this modification improves many published Silks. The second modification is to begin the DAT carpet not on the replicator's own code, but on that of an "uncle" - a paper from a previous generation. My argument for this follows Beppe's reasoning that against scanners and SPL clears it pays to kill off older copies in case they have been stunned. OTOH there is little point a replicator wiping itself because, if it is still capable of wiping, it is probably not damaged. There is something rather special about those step sizes, and I don't know what. I have tried over 60 alternatives, hand-picked from amongst thousands of PMARS simulations -- see Beppe's article in an early Corewarrior for the general approach. None were better and most were clearly worse, so I will just give George's account of how he generated the numbers: well it's kind of a funny story ... if you look at frogz .. that's where CCPaper originated ... Frogs is a simple paper that just happend to have the right step sizes ... and believe me I chose those in random ... The skill of "Just happening" to pick the right numbers is one I *want*. The result is a very aggressive, fast-splitting replicator. It spends a substantial effort on self destruction, hence the name, and this makes it more vulnerable than most Silks. For example it sometimes loses to DAT bombers, and fairly frequently to incendiary bombers (and vampires?). The compensation is that it achieves many more kills than most Silks, especially against traditional scissors such as scanners and SPL clears. This makes it particularly hard for P-spacers to handle as they are often fooled into fielding the wrong warrior. The combination achieved pretty good scores against most warriors, with the worst results coming against imp-heavy warriors that are usually too tough to kill. I suspect that its recent slide has been due to an increase in the population of Oneshots, as these stun quickly but, unlike single-process scanners, are hard to kill by over-writing. There is a small mystery in the hill scores. I have resubmitted this code to Pizza and it barely makes it onto the hill -- yet the original is still in the top half. I know that the starting positions will have changed since I first submitted, but a 15-point difference in scores is rather hard to understand. And the ;name? Spot the classic bug... I didn't. ;redcode-94 test ;name Benj's Revenge 1.0 ;author Robert Macrae ;assert CORESIZE==8000 ;strategy Q^2 (new), and Suicidal Paper ; Name obvious to those who know my son :-) ; (oops -- classic "Name" bug, but I like the new one too:) ORG start ; History ; 2.1 reverts to Probe attack. ; 2.6 puts gap after paper, before bomber. ; 2.9 2.6 is best yet, so drop to 44 scans. ; Time to stop fiddling! ; Q2NEW ; See q2table.pas for the code to get the table values. ; These [9,5,13,1,2,17,16] give up to 28 distinct targets ; with the greatest at only 34. Easier to use than Probe's, ; as no references leak outside table. ; Checked by exec k, exec k+1 and checking bomb fell on scan mark! ; All the scans included as, depending on step, different ones ; will need to be commented out to avoid self-scans. ; ---------------------------------------------- ;-redcode-94 test ;-name Sad 1.0 ;-author Robert Macrae ;-assert CORESIZE==8000 ;-strategy Suicidal Paper ; 1.0 spends a lot of time bombing, often itself. This makes it ; resistant to carpets and scanners, but it may lose to dwarves! ; Constants straight from ccpaper. len EQU 9 fcp EQU 3039 scp EQU 2365 tcp EQU 777 boot spl 1, >-3000 ; 7 processes replace 9 in CCPaper spl 1, <-3200 ; for cost of extra Mov in launcher. mov -1, 0 frog spl @0, -1 mov }-2, >-2 spl @0, -1 spl @0, -1 mov 2, <-fcp+len+1 ; Wipe uncle. jmp -1, <-10 dat <2667, <2667*2 ; ---------------------------------------------- for 21 dat 0,0 rof ; ---------------------------------------------- QS EQU 100 ; Illustration only! QD EQU 4000 ; Ditto. QB EQU (start+14*QS) ; Ditto. CR EQU (fnd-which) datz EQU (table-3) dat 9*QS, 1*QS ; can get 28 values from this table table: dat 5*QS, 2*QS ; dat 13*QS, 17*QS ; P: add.f table,table ; point into table. Nudge with }{>< and djn.f. slow: add.ab *P,fnd ; adds an element A column (usually) fast: add.b @P,@-1 ; adds an element B column (usually) which: sne.i datz,@fnd ; which half of scan hit? add.ab #QD,fnd ; ---------------------------------------------- ; Original Probe attack COUNT EQU 6 GAP EQU 15 REP EQU 6 mov.i qbomb,@fnd fnd: mov.i -GAP/2,@QB ; picks up table as bomb... add.ba fnd,fnd mov.i qbomb,*fnd add.f qinc,fnd mov.i qbomb,@fnd djn.b -3,#REP jmp boot,}-300 qbomb: jmp -200,GAP qinc: dat GAP,-GAP ; ---------------------------------------------- ; 0/1 cycle ; ---------------------------------------------- start: seq.i QB+QS*0,QB+QS*0+QD jmp which, 0 ; 0 seq.i QB+QS*2,QB+QS*2+QD jmp fast, 0 ; E seq.i QB+QS*1,QB+QS*1+QD jmp fast,

P ; F ; ---------------------------------------------- ; 2 cycles ; ---------------------------------------------- seq.i QB+QS*7,QB+QS*7+QD jmp slow, 0 ; BE seq.i QB+QS*6,QB+QS*6+QD jmp slow,

P ; BF seq.i QB+QS*11,QB+QS*11+QD jmp slow, {P ; AE seq.i QB+QS*15,QB+QS*15+QD jmp slow, }P ; CE seq.i QB+QS*10,QB+QS*10+QD djn.f slow, P ; AD seq.i QB+QS*5,QB+QS*5+QD jmp >fast, 0 ; B seq.i QB+QS*9,QB+QS*9+QD jmp >fast, {P ; A seq.i QB+QS*13,QB+QS*13+QD jmp >fast, }P ; C ; ---------------------------------------------- ; 3 cycles ; ---------------------------------------------- seq.i QB+QS*14,QB+QS*14+QD jmp P, 0 ; BBEE seq.i QB+QS*8,QB+QS*8+QD jmp P,

P ; BEF ; KO to avoid self scan! seq.i QB+QS*12,QB+QS*12+QD jmp P, {P ; ADE seq.i QB+QS*32,QB+QS*32+QD jmp P, }P ; CEF seq.i QB+QS*20,QB+QS*20+QD djn.f P, P ; AADD seq.i QB+QS*4,QB+QS*4+QD jmp }slow, 0 ; EE seq.i QB+QS*3,QB+QS*3+QD jmp }slow, {P ; DE seq.i QB+QS*19,QB+QS*19+QD jmp }slow, }P ; FE ; seq.i QB+QS*2,QB+QS*2+QD ; Duplicates a faster scan ; djn.f }slow, P ; DD ; Duplicates a faster scan ; seq.i QB+QS*10,QB+QS*10+QD ; Duplicates a faster scan ; jmp P ; BBG ; seq.i QB+QS*34+CR,QB+QS*34+QD+CR ; jmp }fast, {P ; AAG ; seq.i QB+QS*42+CR,QB+QS*42+QD+CR ; jmp }fast, }P ; CCG ; seq.i QB+QS*52+CR,QB+QS*52+QD+CR ; djn.f }fast, P ; AAAAG ; seq.i QB+QS*16+CR,QB+QS*16+QD+CR ; djn.f }fast, }slow ; G jmp boot ; If you don't spot any, get sad ;-) end ______________________________________________________________________________ Questions? Concerns? Comments? Complaints? Mail them to people who care. authors: Beppe Bezzi or Philip Kendall or Anton Marsden or Christian Schmidt