Collecting data for Hersrch updates

[dvgrn]

This is a thread to collect information on new Herschel tracks. I'll need a whole pile of pattern files and associated statistics to update Karel Suhajda's Hersrch search program to take advantage of the recent deluge of new Herschel conduits.

I'll be doing incremental updates to the data.ini file and conduit library. Contributions are definitely welcome -- just pick a conduit and reply to this message with all the patterns and statistics that Hersrch will need to use it in searches (see below). This adds up to a substantial amount of work, and more hands will make it much lighter!

The latest version of Hersrch can be found here (4 December 2012). Conduits added so far: Bx198, Bx260, Bx285, F114p8, F171, F266(m), Fx266(m), Fx78p8, Fx177p6, L161p8, L198, L255, L309, Lx79p6, Lx138, Lx142, Lx159, Lx198(b), Lx212, Lx216p8, Bx167, R126, R178, R246, Rx140, Rx162, and small improvements to the Fx77SW and Bx222w2 variants. See changelog.txt for details. I update this paragraph for every new incremental release. If you find anything wrong with the current Hersrch, please let me know.

Edit by Sokwe: a small update (15 July 2022) can be found here.

Here's a summary and walkthrough of the information needed for each new conduit:

• 1) RLE of the conduit, with "ghost" input and output Herschels.
  Input Herschel is o$obo$3o$2bo orientation.
  2) RLE of the conduit's active region (where other conduits' catalysts must not overlap).
  The input Herschel is shown as a hole in the region matching the output Herschel.
  3) Statistics for data.ini -- compression, offsets, etc. Details and example below.
  4) List of conduits that can/cannot precede or follow the conduit in question.

As I mentioned, the first thing I've done is to add a known period-6 conduit (by Dietrich Leithner, 1 December 1997). Fx177p6 was only accidentally overlooked when Hersrch was first written. Also, the active reaction features a spontaneously generated eater -- can't just let something like that go to waste...!

Fx177p6.rle:

Code: Select all

#C vfx177p6 standard -- ghost Herschels at start and end points
x = 39, y = 45, rule = B3/S23
14b2o$13bobo$13bo$11b2ob4o$7b2obo3bo2bo$7b2obobo5b2o$10bo4b3o2bo$4b2o
4bob2o4b2obo$4b2o5bo3bobo3bo$4b2o6b3o3b3o$4bo9bobobo$3bobo9b3o12b2o$4b
obob3o5bo14bo$5bob4o20bobo$6bo25bobo$33bo3b2o$37b2o4$38bo$36b3o$36bo$
36bo2$6bo$6bo$6b3o$8bo5$21b5o$3b5o11b3obob3o$b3obob3o8bo4bo4bo$o4bo4bo
3b2obob2obo2b2obo$ob2o2bob2obobobobobo3b2o2bo$bo2b2o3bobob2o2bobo2bo2b
2o$2b2o2bo2bobo5bobo4bo$4bo4bobo6b2ob4o$4b4ob2o9bo$8bo11bobo$6bobo12b
2o$6b2o!

Code: Select all

#C ufx177p6.rle -- active region of standard Fx177p6
x = 36, y = 37, rule = B3/S23
12b3o$12b3o$12b3o$b2o9b3o$4o8b3o$3o9b3o$bo10b3o$obo3bo5b3o$b7o4b3o$2b
6o4b3o5b3o$3bob2o5b3o3b6o$11b4o3b7o$10b6ob8o$10b16o3b3o$10b17ob4o$10b
22o$7b25o3bo$4b28ob3o$4b28obobo$2b30obo$2b22o3b2o2bo$2bob20o$2bobob17o
$2bo3b17o$2b3ob6ob9o$2b7o10b3o$2b3o15b2o$2b2o15b3o$b3o15b3o$b3o15b3o$b
3o15b3o$b3o15b3o$b3o15b3o$b3o15b3o$b3o15b3o$b3o15b3o$b3o!

In Hersrch it's often necessary to provide a separate pair of pattern files for cases where the _next_ conduit suppresses the Herschel's first natural glider (FNG). If the next conduit's FNG can escape without damaging anything, nothing will need to change (e.g., in F117 and many others). But in a fair number of cases, an eater has to be placed to keep that following FNG from hitting the current conduit. If there won't _be_ a following FNG, that eater is unnecessary. Sometimes this can make the difference between the conduit being Spartan or non-Spartan (easily glider-constructible or not).

Code: Select all

#C vfx177p6x.rle -- variant with no tub-with-tail eater needed.
#C The three dots are a visual reminder: no glider from the output H.
x = 39, y = 45, rule = B3/S23
14b2o$13bobo$13bo$11b2ob4o$7b2obo3bo2bo$7b2obobo5b2o$10bo4b3o2bo$4b2o
4bob2o4b2obo$4b2o5bo3bobo3bo$4b2o6b3o3b3o$4bo9bobobo$3bobo9b3o$4bobob
3o5bo$5bob4o$6bo6$38bo$36b3o$36bo$36bo2$6bo$6bo$6b3o$8bo5$21b5o$3b5o
11b3obob3o$b3obob3o8bo4bo4bo$o4bo4bo3b2obob2obo2b2obo$ob2o2bob2obobobo
bobo3b2o2bo$bo2b2o3bobob2o2bobo2bo2b2o$2b2o2bo2bobo5bobo4bo$4bo4bobo6b
2ob4o$4b4ob2o9bo$8bo11bobo$6bobo12b2o$6b2o!

Code: Select all

#C ufx177p6x.rle -- active region of Fx177p6 when followed by
#C  a glider-suppressing "dependent conduit" such as F166
14b2o$13bobo$13bo$11b2ob4o$7b2obo3bo2bo$7b2obobo5b2o$10bo4b3o2bo$4b2o
4bob2o4b2obo$4b2o5bo3bobo3bo$4b2o6b3o3b3o$4bo9bobobo$3bobo9b3o$4bobob
3o5bo$5bob4o$6bo3$36bo2$34bo$38bo$32bo3b3o$36bo$36bo2$6bo$6bo$6b3o$8bo
5$21b5o$3b5o11b3obob3o$b3obob3o8bo4bo4bo$o4bo4bo3b2obob2obo2b2obo$ob2o
2bob2obobobobobo3b2o2bo$bo2b2o3bobob2o2bobo2bo2b2o$2b2o2bo2bobo5bobo4b
o$4bo4bobo6b2ob4o$4b4ob2o9bo$8bo11bobo$6bobo12b2o$6b2o!

Notice that there's not much difference between ufx177p6.rle and ufx177p6x.rle. In this case, there are many types of eaters that would work equally well in various locations to protect the p6 sparker -- a tub-with-tail, 7x9 eater, or a standard fishhook eater welded to the sparker, for example. So there's only a small area where the glider is guaranteed to travel before an eater can get hold of it.

There's no other easily-defined area that's definitely forbidden for other conduits to overlap, because the eater can be placed in a lot of possible places along the glider's diagonal. For all Hersrch knows, something might be made to work in any particular case of overlap in that area. So the general rule is to err on the side of caution in making these maps (as the minimal area reserved for the p6 sparkers also shows!) It's probably better to let users manually reject a few impossible tracks here and there, rather than forbidding too much and missing out on a really nice close custom fit.

To go along with all of this, we have to add a new section to the "Component Definition" section of Hersrch's data.ini file:

Code: Select all

Fx177 Fx 177 30 -6

  # Standard variant
  VFX177P6 6 186 0 GSU data/vfx177p6.rle 7 27 data/ufx177p6.rle 4 23
  
  # Standard variant without eater - no following glider
  VFX177P6X 6 186 0 GSX data/vfx177p6x.rle 7 27 data/ufx177p6x.rle 4 23

• Fx : forward conduit, output is flipped
  177 : time from Herschel to Herschel
  30 -6 : dx, dy offset from old Herschel to new one
  6 : period of the oscillators in the conduit
  186 : minimum number of ticks before the next Herschel can be safely sent
  0 : zero if no maximum compression (this is the normal case)
  GSU : GS = glider suppressed, GR = glider released; U = uses following glider,
  ..... : X = glider must not be released from the following Herschel
  7 27 : dx, dy from LOWER left corner of vfx177p6x.rle to center cell of Herschel
  4 23 : dx, dy from LOWER left corner of ufx177p6x.rle to center cell of Herschel

You can read those last dx, dy coordinates straight out of Golly, like this: copy the pattern in question, choose File > Open from Clipboard in Golly, hit "Y" to mirror the pattern vertically, then point to the central ON cell in the Herschel heptomino and look at Golly's cursor coordinates.

The only other statistics that have to be collected for each conduit are the forbidden variant pairs -- which conduits should Hersrch not even bother to try to connect to each other? Having this list up-to-date saves Hersrch a lot of time during searches, and/or saves searchers a lot of time throwing out unworkable results. I'll post my working pattern for this task in a separate message.

[Sokwe]

A while back I added Lx138 to look for gun reductions.

Code: Select all

Lx138 Lx 138 17 -32

  # Standard variant, with eater
  VLX138 1 84 0 GRU data/vlx138.rle 2 4 data/ulx138.rle 2 2

  # Standard variant, without eater (no glider)
  VLX138X 1 84 0 GRX data/vlx138x.rle 2 4 data/ulx138x.rle 2 2

vlx138.rle:

Code: Select all

18b2o$19bo$19b3o6$12b2o$11bobo$11bo$10b2o7$20b2o$9bo10b2o$9b3o$o11bo
14bob2o$3o8b2o14b2obo$3bo$2b2o3$24b2o$24b2o3$bo$bo22b2obo$b3o20b2ob3o$
3bo11b2o13bo$15bo8b2ob3o$16b3o4bo2b2o$18bo4b2o!

ulx138.rle:

Code: Select all

18b3o$19bo$19b3o2$16b7o$15b8o$14b8o$14b8o$12b3o2b6o$13b2ob7o$16b6o$16b
6o$16b6o$15b8o$16b8o$15b9o$15b9o$15b10o$15b10o$15b11o$15b11o$12bo2b13o
$11b2o4b8o2b2o$3bo7b13o$2b2o3bo5b10o$2b8o2b11o$4b19o$4b22o$3b23o$b25o$
24o$ob14o4b4o$obob4ob6o6bob3o$o3b4o2bob5o6b4o$3ob4o7b2o8bo$7o8bo!

vlx138x.rle:

Code: Select all

18b2o$19bo$19b3o16$20b2o$9bo10b2o$9b3o$o11bo14bob2o$3o8b2o14b2obo$3bo$
2b2o3$24b2o$24b2o3$bo$bo22b2obo$b3o20b2ob3o$3bo11b2o13bo$15bo8b2ob3o$
16b3o4bo2b2o$18bo4b2o!

ulx138x.rle:

Code: Select all

18b3o$19bo$19b3o2$17b6o$17b6o$17b5o$16b6o$17b6o$16b7o$16b6o$16b6o$16b
6o$15b8o$16b8o$15b9o$15b9o$15b10o$15b10o$15b11o$15b11o$12bo2b13o$11b2o
4b8o2b2o$3bo7b13o$2b2o3bo5b10o$2b8o2b11o$4b19o$4b22o$3b23o$b25o$24o$ob
14o4b4o$obob4ob6o6bob3o$o3b4o2bob5o6b4o$3ob4o7b2o8bo$7o8bo!

The information to add to [pairs] is the following:

Code: Select all

VBX222-VLX138
VBX222-VLX138X

VBX222W1+VLX138
VBX222W1+VLX138X

VL214P8-VLX138
VL214P8-VLX138X

VLX65P8B-VLX138
VLX65P8B-VLX138X

VLX65P8X1-VLX138
VLX65P8X1-VLX138X

VLX65P8-VLX138
VLX65P8-VLX138X

I have not made any attempt to record impossible sequences.

@Dave
I edited your first post to fix a couple of links. Also, after the stable conduits are added, it might be good to add some of the new periodic conduits. I posted one type here, and "Extrementhusiast" posted a fast p6 conduit here. The unix in this last conduit can be replaced with a pipsquirter to allow common eater placements. The larger p6 can likely be modified to give more room for conduit connections as well.

[dvgrn]

A while back I added Lx138 to look for gun reductions.
...
I edited your first post to fix a couple of links. Also, after the stable conduits are added, it might be good to add some of the new periodic conduits. I posted one type here...

Looks good -- thanks! I certainly have no objection to adding compact periodic conduits. In fact my baseline version of Hersrch contains not only Fx177p6 but also a fast and slow version of Fx78p8, a truncation of F117 that uses the output Herschel's following glider. A number of recent conduits include similar cleanups.

Your period-8 R135 conduit allows for an Rx96p8 version along the same lines, by the way (when this trick works, it always flips the Herschel and cuts off 39 ticks):

Code: Select all

#C slow and fast Rx96p8 conduits based on Sokwe's R135
x = 64, y = 38, rule = LifeHistory
6.2A37.2A12.2A$6.A.A.2A.A31.A.A.2A.A5.3B$8.A.A.2A33.A.A.2A5.A3BA$7.B
2A36.B2A8.A4BA$8.2B6.2A29.2B7.A.ABAB$6.4B6.3B26.4B5.BABA.A$6.6B2.A3BA
26.6B2.A4BA$5.9BA4BA24.9BA3BA$4.9B2.BABA.A22.9B3.3B$4.9B4.A.ABAB20.9B
3.2A$2.11B5.A4BA17.11B$2.11B6.A3BA17.11B$2.11B6.3B19.11B$3.10B.B5.2A
20.10B.B$3.11B2A26.11B2A$2.10B.B2A25.10B.B2A5.AB$.10B3.B5.2B18.11B2.B
5.B2A$.10B8.3B18.9B.B8.2ABAB$2.9B8.3A2B17.8B9.A2B3A$2.8B8.B3A2B17.8B
8.A.ABAB$3.2B2A3B7.B.3AB19.2B2A3B6.BABA.A$2.3B2A3B5.B3A.B20.3B2A3B5.
3A2BA$2.9B3.2B3AB21.8B4.2BAB2A$2.10B.3B3A22.8B7.2AB$.12B3.3B21.9B4.AB
.BA$2.11BAB.2B23.8B.2BABAB$3.9BABAB26.10B2A2B$3.9B2A2B26.7B.4B.B2A$2.
12B.B2A23.8B.2B3.BA.A$2.10B3.BA.A22.8B.B2A5.A$2.6B.2B2A5.A22.6B.B2.2A
5.2A$.2B3C2B3.2A5.2A20.2B3C2B$.4BC2B32.4BC2B$4B3CB31.4B3CB$8B31.8B$8B
31.8B$8B31.8B$7B32.7B!

On its first pass, Hersrch discards conduits that don't match a given search's required period, so there's very little cost to adding periodic conduits. The most painful part is generating all the statistics and pattern files! I'll probably have a Python script out fairly soon to make the data collecting a bit easier... no promises, though, so if someone else writes one first, I'll be most grateful.

-- Now, I'm really most interested in the stable conduits, myself, and especially the Spartan ones. All my recent projects seem to involve self-constructing circuitry... so I'll probably avoid cataloguing any periodic conduits from now on. I have the feeling that currently known periodic circuitry is only the tip of the iceberg, and if we start looking too hard for oscillator-assisted Herschel tracks, we might really get buried! There are relatively few Life problems that periodic circuitry can solve, so I'd rather focus on the P1 stuff.

However, if complete statistics and pattern files for a compact periodic conduit show up on this thread, I'll be happy to roll them in to an incremental update, even if they're out of order in terms of (my) priorities.

I might make an executive decision sometimes to leave a conduit out, if it seems just too bulky or too slow to recover. Over 512 ticks seems definitely too slow, for example, unless it's something unusual like a compact Spartan Herschel splitter.

If the oscillators are huge and get in the way of making connections to most other conduits, that's a problem too. There are already a few registered conduits along these lines, and I'll be reluctant to add very many more of them. As long as the statistics are published here, anyone who disagrees with my executive decisions can easily edit their data.ini and build their own custom Hersrch...!

[Sokwe]

7 27 : dx, dy from LOWER left corner of vfx177p6x.rle to center cell of Herschel
4 23 : dx, dy from LOWER left corner of ufx177p6x.rle to center cell of Herschel

You seem to have done this from the upper left corner.

Also, Here is another conduit. It's relatively fast but not very compact.

Code: Select all

Lx212 Lx 212 31 -41

  # Standard variant, with eater
  VLX212 1 92 0 GRU data/vlx212.rle 3 6 data/ulx212.rle 2 3

  # Standard variant, without eater (no glider)
  VLX212X 1 92 0 GRX data/vlx212x.rle 3 6 data/ulx212x.rle 2 3

vlx212.rle:

Code: Select all

33b2o$34bo$34b3o6$27b2o$26bobo$26bo$25b2o$33b2o$33b2o2$20b2o$10bo10bo
20b2o$10b3o8bobo18b2o$13bo8b2o$12b2o3$41b2o$41bo$42b3o$44bo5$2o$bo32b
2o$bobo30bo$2b2o31b3o$37bo6$2bo$2bo22b2o$2b3o20bo$4bo10b2o3b2o4b3o$13b
o2bo4bo6bo$13b3o4bo$20b2o$13b2obo$13bob2o!

ulx212.rle:

Code: Select all

32b3o$33bo$33b3o2$30b7o$29b8o$28b8o$28b8o$26b3o2b6o$27b2ob7o$30b6o$30b
6o$31b6o$30b9o$30b9o$31b9obo$29b14o$22b2o4b15o$12bo8b3ob17o$11b2o10b
19o$11b4o2b3o3b17o$13b21obob5o$11b25o3b3o$11b25o4bo$10b26o$9b26o$9b25o
$10b17o3b4o$10b16o4b4o$7b14obo8b3o$7b15o9b4o$6b15o11b3o$2b2o2b15o12bo$
b3ob15o$3b20o$3b20o$4b20o$3b22o$b24o$24o$ob13ob10o$obob4ob7o2b2o4b2o$o
3b4o2b3obo4bo4bo$3ob4o4b4o3b2o$7o4b2o2bo4bo$12b3o!

vlx212x.rle:

Code: Select all

33b2o$34bo$34b3o10$33b2o$33b2o2$20b2o$10bo10bo20b2o$10b3o8bobo18b2o$
13bo8b2o$12b2o3$41b2o$41bo$42b3o$44bo5$2o$bo32b2o$bobo30bo$2b2o31b3o$
37bo6$2bo$2bo22b2o$2b3o20bo$4bo10b2o3b2o4b3o$13bo2bo4bo6bo$13b3o4bo$
20b2o$13b2obo$13bob2o!

ulx212x.rle:

Code: Select all

32b3o$33bo$33b3o2$31b6o$31b6o$31b5o$30b6o$31b6o$30b7o$30b6o$30b6o$31b
6o$30b9o$30b9o$31b9obo$29b14o$22b2o4b15o$12bo8b3ob17o$11b2o10b19o$11b
4o2b3o3b17o$13b21obob5o$11b25o3b3o$11b25o4bo$10b26o$9b26o$9b25o$10b17o
3b4o$10b16o4b4o$7b14obo8b3o$7b15o9b4o$6b15o11b3o$2b2o2b15o12bo$b3ob15o
$3b20o$3b20o$4b20o$3b22o$b24o$24o$ob13ob10o$obob4ob7o2b2o4b2o$o3b4o2b
3obo4bo4bo$3ob4o4b4o3b2o$7o4b2o2bo4bo$12b3o!

pairs data:

Code: Select all

VBX222-VLX212
VBX222-VLX212X

VBX222W2+VLX212
VBX222W2+VLX212X

VFX77S-VLX212
VFX77S-VLX212X

VFX77SW+VLX212
VFX77SW+VLX212X

VL214P8-VLX212
VL214P8-VLX212X

VLX65P8B-VLX212
VLX65P8B-VLX212X

VLX65P8X1-VLX212
VLX65P8X1-VLX212X

VLX65P8-VLX212
VLX65P8-VLX212X

@Dave
What conduits do you intend to add first? I do not want to add any conduits that you're already working on.

[dvgrn]

7 27 : dx, dy from LOWER left corner of vfx177p6x.rle to center cell of Herschel
4 23 : dx, dy from LOWER left corner of ufx177p6x.rle to center cell of Herschel

You seem to have done this from the upper left corner.

Huh, that's puzzling -- I remember carefully inverting the pattern before getting the coordinates, but now it sure looks like the numbers should be 7 17 and 4 13, as you say. The really strange thing is that I ran several tests with the updated Hersrch, to make sure that it could build tracks with the new conduit included...

-- Ah, wait, no! I only did two searches, one to find a track exactly the same length as Fx177p6x, and one to complete a loop from the end of Fx177p6x to the beginning. So Hersrch never actually connected the beginning or end of the new conduit to anything else, and I never saw the mismatch... Well, no harm done, it's an easy fix. From now on I'll do a less silly set of tests.

Also, Here is another conduit. It's relatively fast but not very compact.

Very nice! Anything that recovers in under 100 ticks is blazingly fast, as far as I'm concerned, and the size is perfectly reasonable for a 212-tick delay. I'll roll it right into the next update.

What conduits do you intend to add first? I do not want to add any conduits that you're already working on.

Well, I've just finished getting familiar with Guam's constructions (see next posting) so I haven't gotten back to work on Hersrch statistics in the last week. My current plan is to add all of Guam's Spartan conduits first, then the rest of his collection.

-- Not that I want to discriminate against your own equally interesting inventions... but my particular reason for updating Hersrch is to simplify self-constructing circuitry, whereas you seem to favor your distinctly non-Spartan BoS catalysts (Bookend on Snake).

So if it's okay, I'll leave it up to you for now to collect statistics on your Bx167, Bx285, L198, L255, Lx159, Lx496, and the periodic R135 variants, and see if I can get through at least some of Guam's collection before you're done with those...!

By the way, does anyone have a good stamp-collection pattern with history cells for the new and old Herschel conduits? If not, I'm about to build one, to simplify the process of figuring out forbidden and allowed pairs as each new conduit is added.

[dvgrn]

As my last post indicates, I've been working on understanding some of the more complicated new Herschel tracks that have showed up on the conwaylife forums in the last year, with an eye to improving Hersrch's conduit database.

Many of the new conduits have significant restrictions on what other conduits can precede or follow them, so I found it useful to have working examples of each conduit to make sure I wasn't missing any important details. Here's a sample pattern that strings together many of Guam's discoveries from the past year.

I have failed to include a number of other interesting new conduits, by Sokwe and Calcyman for example, but there are several blocked-off Herschel outputs that could easily be extended if someone wants to attempt it... Eventually I'd like to see a much more comprehensive version of Golly's old track.py that showcases a really wide variety of signal-processing circuitry.

Code: Select all

#C Some of Guam's stable circuitry, 2011-12
#C See postings by Guam, Sokwe, and Calcyman,
#C http://www.conwaylife.com/forums/viewtopic.php?f=2&t=279
#C http://www.conwaylife.com/forums/viewtopic.php?f=2&t=148
x = 557, y = 388, rule = B3/S23
383b2o$383b2o5b2o$390b2o2$346b2o16bo$346b2o16b3o21b2o$367bo20b2o$341b
2o23b2o26b2o$341b2o51b2o8$334b2o$333bobo$333bo$332b2o25bo$358bobo$358b
2o6b2o$367bo$343b2o19b3o$344bo8b2o9bo$344b3o6bo$354b3o28b2o$356bo28b2o
4$337b2o$336bobo$336bo31b2o$335b2o32bo$369bobo$370b2o5$345b2o$345b2o2$
352bo$352b3o11b2o$335b2o18bo9bobo$335b2o17b2o9bo$325b2o37b2o$326bo47b
2o17b2o$326bobo45b2o18bo$327b2o62b3o$391bo2$384b2o$384b2o5$331b2o$330b
obo$330bo43b2o$329b2o44bo$375bobo$376b2o39b2o$417bobo$419bo$419b2o$
270bo20bo15bo9bo28b2o47bo$269bobo7bo11b3o13b3o5b3o28b2o45b3o$262b2o6bo
6b3o14bo15bo3bo68b3o6bo$262b2o12bo16b2o14b2o3b2o9bo57bo8b2o$250b2o24b
2o47b3o54b2o$250bo77bo$251b3o73b2o$253bo$302bo$256b2o42b3o$256b2o25b2o
15bo96b2o$283b2o15bo96bo$245b2o148bobo$245b2o148b2o3$381b2o$280b2o98bo
bo$280bo19b2o78bo$281b3o15bobo11b2o6bo33b2o22b2o17b2o23bo$261b2o20bo
15bo13b2o6b3o31b2o41bo22b3o$261b2o35b2o24bo3b2o19b2o45bobo21bo$323b2o
3bo20b2o29b2o14b2o22b2o$329b3o47bobo26bo$331bo47bo28b3o$351b2o25b2o31b
o$344b2o5b2o57b2o40bo$344b2o106b3o$455bo$424b2o11bo16b2o$424b2o11b3o$
440bo$439b2o3$463b2o$463b2o$448bo$375b2o69b3o$375b2o52b2o15bo$401b2o
26b2o15bo19b2o$401b2o63b2o2$380b2o$380b2o$376b2o$376b2o48b2o$227b2o
197bo19b2o$228bo198b3o15bobo$225b3o154b2o45bo15bo$225bo156b2o60b2o2$
450b2o$450bobo$451bo4$447b2o17b2o$406bo41bo17bo$405bobo40bobo13bobo$
406bobo2b6o26bo5b2o13b2o$407b2o5bo2bo23b3o$414bo2bo22bo$415b2o23b2o$
471bo$414b2obo51b3o$413bobobo50bo$413bobobo50b2o$414b3o2$413bo3bo$414b
obo$415bo$411b7o2$414bo2bo30b2o$414bo2bo30bo$415b2o32b3o$451bo$415b2o$
414bo2bo$414bo2bo26b2o$415b2o27bobo13bo$446bo12bobo$247bo163b7o28b2o
11bobo$246bobo211bo$245bobo167b2o$245b2o167bo2bo25b2o$414bo2bo25bo10b
2o$411b6o27b3o6bobo$217bo4bo10bo12bo199bo6b2o$217bo4bo10bo12bo191b2o$
217bo4bo10bo2bo2b2o5bo191bo$217b3o2bo2b2o3b2obobo2bo2bo2b3o192b3o$217b
o2bobobo2bobo3b2o3b4obo2bo194bo$217bo2bobobo2bobo3bobo2bo4bo2bo$218b2o
2bo2b2o3b2obo2bo2b3o2b2o5$212b2o$213bo$81bo131bobo$81b3o130b2o$84bo
125b2o$83b2o126bo$211bobo$207bo4b2o$207b3o31bo230b2o$210bo29bobo229bob
o$209bo30bobo231bo5b2o3b2o$208bo32bo232b2o4bobobobo$208b2o272bobo$480b
obobobo$480b2o3b2o3$11b2o$11b2o42b2o$56bo14bo323b2o$55bo15b3o322bo$5b
2o48b2o17bo153b2o163b3o84b2o$5b2o66b2o152bo2bo162bo86bo$9b2o217bobo19b
2o217b2o7b3o$9b2o218bo20b2o216bobo$468bo$30b2o168b2o30b2o233b2o$30b2o
26b2o140b2o29bobo$4b2o52b2o144b2o25bo$4b2o198b2o24b2o324bo$218b2o41b2o
291b3o$218b2o40bo2bo2b2o285bo$260bobo4bo285b2o$241b2o18bo5bob2o$240bob
o21b2obobo$68b2o44b2o124bo23bo2bo2bo$69bo43bo2bo104bo17b2o20bo4bo2b2o$
53b2o11b3o44bob2o103bobo7b2o29b5o$53b2o11bo44bobobo23b2o79bobo7b2o$
107b2o2b2o2bo23bo68b2o11bo41b2obo211b2o$107b2o6b3o22b3o66bo53bob2o211b
2o$39b2o77bo23bo63b3o$40bo43b2o11bo19b2o87bo262b2obo$37b3o39b2o3b2o9b
3o112b2o257bob2o$37bo15b2o24b2o14bo54b2o58b2o$49b2o2b2o40bo53bobo$10b
2o37bo100bo$11bo38b5o23b2o408b2o$11bobo41bo23bo4b2o401bobo$7bo4b2o30bo
5b2o2b2o20b3o5b2o55bo345bo$7b3o32b3o5b2o24bo63bobo343b2o$10bo30bo98bob
o5b2o$9bo31b2o52b2o44bo6b2o$8bo45bo39bobo$8b2o43bobo38bo372b2o$29bo24b
2o10b2o25b2o373bo$28bobo35b2o400bobo$29b2o19b2o49b2o366b2o15b2o$42b2o
6b2o50bo44bo338b2o$42b2o55b3o44bobo$99bo7b2o38bo$33b2o72b2o360b2o$32bo
bo24b2o407bobo$32bo20b2o4bo408bo61b2o$31b2o20bo6b3o404b2o61bo$54b3o5bo
81b2o249b2o125b2o4bobo$56bo87bobo249bo124bo2bo3b2o$146bo249bobo122bobo
$2o144b2o249b2o114b2o7bo$2o126bob2o214bo166b2o$4b2o122b2obo214b3o136b
2o$4b2o343bo135bo$18b2o117b2o209b2o7b2o127b3o42b2o$18b2o117b2o218b2o
129bo42b2o4$378b2o89b2o$97b2o279b2o88bobo$30b2o66bo369bo52b2o$30b2o63b
3o49b2o266b2o50b2o52b2o$8b2o85bo51bo240bo26bo$9bo135bobo240bo24bobo$6b
3o136b2o241b3o22b2o$6bo383bo3$477b2o$127b2o348b2o$128bo259b2o$15b2o7b
2o34b2o66bobo257b2o94bob2o$16bo7bobo32bobo67b2o353b2obo$13b3o10bo32b2o
$13bo12b2o305b2obo78b2o$333bob2o78b2o49b2o$54bo412bo$53bobo24b2o282b2o
101bobo$53bobo24bo284bo26b2o74b2o$54bo23bobo209b2o70b3o26bobo11b2o$78b
2o210b2o70bo28bo13bo$61b2o246b2o79b2o14b3o73b2o$62bo246bo26b2o70bo73bo
bo$62bobo242bobo25bobo70bob2o72bo$63b2o65b2o171b2o2b2o22b2o2bo31b2o37b
3ob2o72b2o$130b2o14b2o128b2o25b2o25bob4o32bo36bo$122b2o22bo130bo52bo
26b2o6b3o9b2o25bobo22b2o$123bo23b3o127bobo51b3o23bobo5bo11bo8bo18bo23b
2o$123bobo23bo128b2o54bo24bo18b3o5b3o$124b2o182b2o23b2o24b2o19bo8bo$
308b2o78b2o13b2o64b2o$80b2o321b2o65bo$79bo2bo251b2o19b2o98b2o13b3o11b
2o$64bo15b2o61bo190b2o19b2o98bobo26b2o$45b2o17bobo74b3o312bo$46bo17b3o
64b3o6bo137bo214b2o$43b3o20bo64bo8b2o136bo213bobo$33b2o8bo41b2o43b2o
130b2o3b2o9b3o102b2o107b2o$33bobo49bobo174b2o3b2o11bo71b2o30bo51b2o$
34b2o41bo8b2o59b2o203bobo29bobo49b2o$75b3o69bo142b2o62bo30b2o$74bo70bo
bo142b2o2b2o58b2o131b2o$74b2o69b2o147bobo189bo2bo$39b2o255bo40b2o60b2o
86b2o$38bo2bo254b2o40bo8b2o50bobo$39b2o297b3o6bo53bo$259b2o87b3o50b2o
20b2o$127b2o19b2o109b2o13b2o74bo72b2o$127b2o19b2o124b2o51b2o98b2o$327b
2o52bob2o41bobo$381b2obo41bo34b2o$126b2o29b2o138b2ob2o123b2o35bo$56b2o
69bo29bo140bobo89b2o46b2o19b3o$56bobo65b3o28bobo139bo2bobo87b2o46b2o
19bo$58bo64bo31b2o140b2o2b2o$58b2o63bob4o208b2o131b2o$41b2o35b2o44b2o
2bo208b2o101b2o28b2o$40bobo23bo11bo49bobo127b2o180b2o$40bo24bobo11b3o
47b2o127b2o39b2obo$39b2o24bobo13bo217bob2o$66bo270b2o$337b2o$125bo$60b
2o62bobo176b2o23b2o$59bobo63bo176bobo12b2o9b2o$59b2o241bo14b2o$293b2o
5b3o$293b2o4bo$299b2o$149b2o$149b2o$259b2o37b2o$260bo37b2o$257b3o4b2o
3b2o$124b2o131bo6bo4bo2bo$123bobo139bo4b3o143bo$123bo21b2o117b2o32b2o
115bobo$122b2o21bo123bob2o7b2o3bo12b2o114bobo$146b3o120b2obo7bo3bobo
127b2o$148bo132bo3b2o14b2o$282bo18bobo$280bob5o16bo82bo4bo10bo12bo$
280b2o4bo16b2o81bo4bo10bo12bo$283b3o13b2o85bo4bo10bo2bo2b2o5bo$280b2ob
o15bobo84b3o2bo2b2o3b2obobo2bo2bo2b3o$280bobo18bo84bo2bobobo2bobo3b2o
3b4obo2bo$301b2o83bo2bobobo2bobo3bobo2bo4bo2bo$387b2o2bo2b2o3b2obo2bo
2b3o2b2o$125bo$125b3o159bo$128bo157bobo$127b2o157b2o$135b2o$135b2o3$
195b2o$166bo28bo$151b2o13b3o23b2obo$151bo17bo23bobobo$120bo31bo15b2o
23bo2b2o$120b3o28b2o38bobo15b2o$123bo63b2o2b2o16b2o$122b2o63b2o29b2o
81b2o$218bo82b2o$177bo38bobo$175b3o38b2o$158b2o15bo81b2o$158b2o15bo82b
o$257bo$119b2o136b2o$118bobo$118bo$117b2o$155b2o$155bo19b2o55b2o$124b
2o3b2o8b2o15bo17bobo55b2o26b2o15bo$124b2o3b2o9bo14b2o17bo85b2o15bo22b
2o$137b3o33b2o102b3o20bo$137bo141bo10b2o3b2o4b3o$181b2o105bo2bo4bo6bo$
182bo14b2o89b3o4bo$179b3o16bo96b2o$179bo7b2o6b3o32b2o56b2obo$187b2o6bo
33bobo38b2o16bob2o$203bo25bo41bo$202bobo9b2o12b2o25b2o11b3o$203bo10bo
40b2o11bo$215b3o$217bo$241b2o$242bo$239b3o$239bo$251b2o$251bo$252b3o$
254bo!

Many of Guam's new mechanisms use standard blocks and gliders from B heptominos and Herschels to clean up extra gliders in interesting new ways. The pattern includes a 'Spartan' section at the top and right. Ghost Herschels mark the beginning of each conduit.

I'll make a LifeHistory version of this pattern at some point, with labels for each conduit included in the pattern itself. Meanwhile, here are some notes to help make sense of it all (I hope):

Code: Select all

T = 0:
p504 gun (skomick, 30 Dec 2011) based on R126 by Guam, 28 Dec 2011

T = 330:
Guam's composite glider-to-Herschel converter, 444-tick recovery

T = 872:
Lx309 by Guam, 7 Oct 2011, followed by an arbitrary Fx77 spacer

T = 1258:
Fx266 (H-pi-H by Guam, 7 Oct 2011) + Lx200

T = 1724:
F266 (pi can be catalyzed in two mirror-symmetric ways) + F166

T = 2156:
Lx159 (by Sokwe, 8 September 2012)

T = 2315:
Lx198 by Guam, 19 Oct 2012, using Bx167 by MM, 8 Sep 2012

T = 2513:
Lx291 (H-pi-H + Bx167) Herschel splitter by Guam, 18 Oct 2012
... This conduit emits G2 gliders plus the standard H output.
... G2 = two parallel gliders separated by 2 cells.
... These can be accepted by a Callahan/Silver Herschel receiver.

T = 2804:
Lx496 (found by Sokwe, improved by Guam 27 Oct 2011)

T = 3000:
Trigger glider reaches sample Herschel receiver (center of pattern).
... Several new receiver variants have been found recently.
... This one is by Guam, 13 October 2012, and Calcyman.
... Here the Herschel output is 'blocked' with a beehive.

T = 3300:
Rx140 by Guam, 13 Oct 2012, improved by Sokwe, + F117.
... First block-suppressing eater in F117 must be removed.

T = 3519:
First glider hits the required standard Herschel receiver
... no other conduit can supply input to the Spartan R126 H-splitter.

T = 3712:
R126 variant: emits G4 gliders along with the standard H output.
... The conduit after R126 must omit the initial eater.


T = 4293:
First glider hits 2G -> G adjustable reflector (Guam, 8 Oct 2011).
... The G5 H-to-2-glider converter is not Spartan, unfortunately --
... no really compact Spartan G2/G5/G6 has been discovered yet.

T = 4580:
First glider hits new Spartan G4 receiver (Guam, 28 Dec 2011).

Spartan section -- glider-constructible still lifes, 7 cells or less:

T = 4865:
R178 = F114 + R64 + dependent conduit (F166 in this case)
... F114 = Fx153 reduced by 39 ticks, block+snake catalyst removed.

T = 5209:
Bx198 = Lx134 + R64 + dependent conduit (F166)
... Lx134: first two catalysts by David Buckingham, 15 Nov 1992,
... next four catalysts by Dietrich Leithner, 27 Apr 1997,
... last two catalysts suppress pi:  Guam, 28 Dec 2011)

T = 5573:
Lx142 = Fx78 + R64 + dependent conduit, here F166
... Fx78 = F117 without the block+snake catalyst.
... A periodic (p8) version of Fx78 was previously known.
... In this variant the dependent-conduit glider does the cleanup.

T = 5881:
Bx260 = Lx134 + R126 + Brice Due's F171.  This branch could continue.
... After F171, the output Herschel is blocked off with a beehive.

[Sokwe]

you seem to favor your distinctly non-Spartan BoS catalysts (Bookend on Snake).

I cannot take credit for this reaction. I was inspired to add the bookend catalyst to ptbsearch based on a vague recollection of a reaction by Brice Due that used it. After running it through ptbsearch, I quickly found the bookend+snake: H --> R+junk reaction, and I just added an eater and block by hand to do the extra cleanup. A few days ago I sought out Brice Due's original message, and found that the bookend reaction that he found was the same one I had rediscovered for the H --> R converter. I'm quite surprised that Brice did not find the full converter, as it only really took a 2-catalyst search.

For Hersrch updates, I will continue to avoid the spartan conduits (mostly because I don't know what extra data they need). I generally intend to add faster conduits first, so the next two that I intend to work on are L198 and Bx167.

More musings on stable technology searches:
There is a ship+eater reaction that occurs surprisingly frequently in ptbsearch output:

Code: Select all

*......
***....
...*...
..**...
.......
.....**
....*.*
....**.
.......
...*...
.**....
.*.....
.*.....

There are more complex ways that a pattern can react with this constellation, such as these by Brice Due:

Code: Select all

x = 46, y = 28, rule = B3/S23
9bo29bo$9b3o27b3o$12bo29bo$11b2o28b2o2$14b2o28b2o$13bobo27bobo$13b2o
28b2o6$o$obo$3o$2bo8$29bo$29bobo$29b3o$31bo!

This is just common enough that I am inclined to add it to ptbsearch as an individual catalyst. Since it is made up of two spartan still lifes, I would hope that it would be easily constructable, but the still lifes are so close together that synthesizing all eight orientations might be difficult.

[Sokwe]

Here's the data for L198:

Code: Select all

L198 L 198 15 -57

  # Standard variant
  VL198 1 94 0 GR data/vl198.rle 1 6 data/ul198.rle 2 3

vl198.rle:

Code: Select all

16b2o$16bo$14b3o16$14b2o$14b2o2$5b2obo$5bob2o7$10bo$9bobo$9b2o12$24b2o
$24b2o12$o$o22b2o$3o20bo$2bo10b2o3b2o4b3o$11bo2bo4bo6bo$11b3o4bo$18b2o
$11b2obo$11bob2o!

ul198.rle:

Code: Select all

16b3o$17bo$15b3o2$14b6o$14b6o$15b5o$15b6o$14b6o$14b7o$15b6o$15b6o$15b
6o$14b8o$13b8o$13b9o$13b9o$12b10o$12b10o$11b11o$11b11o$9b13o$9bo2b8o$
13b7o$14b6o$14b6o$13b7o$14b8o$14b8o$11b2ob9o$12b11o$13b7ob3o$15b10o$
15b10o$14b11o$15b10o$14bob8o$13b9o$11b11o$11b10o$10b12o$9b14o$9b17o$
10b17o$10b17o$9b17o$6b16o$5b16o$5b16o$4b16o$3b20o$3b20o$4b20o$3b22o$b
24o$24o$ob13ob10o$obob4ob7o2b2o4b2o$o3b4o2b3obo4bo4bo$3ob4o4b4o3b2o$7o
4b2o2bo$12b3o!

I don't think any pairs data needs to be added for this conduit.

[dvgrn]

A few days ago I sought out Brice Due's original message, and found that the bookend reaction that he found was the same one I had rediscovered for the H --> R converter. I'm quite surprised that Brice did not find the full converter, as it only really took a 2-catalyst search.

I believe the explanation is that Brice's search program was always at the 'alpha' stage. The F171 was found on maybe the only long test run that Brice tried, with a one-catalyst search! (That explains all the eaters, doesn't it?)

Brice had plans for a search engine that could recognize interesting active objects in any tick of a reaction, rather than just the settled ash after a catalysis was complete. But finishing search programs is a tricky business. I could be wrong, but I seem to recall Brice's F171-finding program never got past the stage where you'd have to change the code to do a different search. So it never got to the point where it was ready for general distribution.

For Hersrch updates, I will continue to avoid the spartan conduits (mostly because I don't know what extra data they need).

They don't need any extra data. The only difference is that the data goes in two places -- into the Spartan data.ini as well as the standard one. Most forbidden-pair information isn't needed for the Spartan version, because the other half of the pair isn't Spartan anyway... but otherwise it's just a straight copy-paste. So if you happen to see any new Spartan conduits lying around that I'm not working on already, by all means send in the information...!

Eventually maybe I'll add a switch in Hersrch to allow searches to be limited to a chosen subset of the conduits. But maintaining a separate data.ini has the great virtue of simplicity, so I'm sticking with that for now! There are minor differences between the conduit pattern files, especially for Fx119, so it's not really just a matter of excluding some conduits.

I've now registered two of Guam's five Spartan conduits: R178 and Lx142. Next up are Bx198, Bx260, and R126.

Thanks for the data on L198 and Lx212! I'll upload an incremental "build" of Hersrch with the four new conduits I have, plus a few other very minor fixes (noted in the change log). But first I have to find time to run a test search or two. Looks like I'm going to be busy or out of town for the next several days -- probably won't get back to registering the rest of the Spartan stuff until next week.

[Sokwe]

I've noticed that a few conduit connections could be "improved" slightly. I'm not sure if these are in any way useful, however.

The large still life in Bx222w2 can be replaced by a block and an eater, as long as the eater from the following conduit is removed (shown here followed by Fx153):

Code: Select all

x = 90, y = 37, rule = LifeHistory
26.2A28.2A$26.A29.A$27.3A6.2A16.A.A12.2A$29.A6.2A16.2A13.A$70.A$47.2A
20.2A$47.2A29.A6.A$76.3A5.A.A$C74.A8.A.A$3C72.2A6.2A.3A$D.C15.2A69.A$
2.C15.2A30.C15.2A15.2A.3A$48.D.C15.2A15.2A.A$48.3C$48.C$52.A$54.A$21.
2A65.2A$22.A32.A32.A$21.A13.2A3.2A15.A28.A.A$21.2A13.A3.A45.2A$33.3A
5.3A$33.A9.A5$54.E$54.E.E$54.3E$56.E20.2A$77.A$66.2A10.A$66.2A2.2A5.
2A$70.A.A$72.A$72.2A!

This slightly reduces the bounding box, but I'm not sure if it would actually reduce the size of any interesting tracks.

Similarly, the eater2/fish hook weld in Fx77sw can be replaced with a larger still life that makes the conduit 1 cell shorter (shown here followed by Fx153):

Code: Select all

x = 77, y = 26, rule = LifeHistory
33.A9.A$33.3A5.3A$36.A3.A13.2A$35.2A3.2A13.A$54.A$A53.2A$3A$3.A$2.2A
11.2A11.C$15.2A9.3C$26.C.D$26.C30.2A15.C$57.2A15.C.D$74.3C$76.C$.E$.E
.E17.2A.2A$.3E18.A.A$3.E18.A.A$21.2A.A.2A11.2A6.A$18.2A2.A.A.A.A10.2A
6.3A$18.2A.A2.A2.2A21.A$21.A.2A24.2A$18.2A.A.A$12.2A4.A2.A.A$12.2A6.
2A.2A!

Again, the eater in the following conduit must be removed.

Edit: The first one actually only needs two blocks, rather than a block and an eater:

Code: Select all

x = 90, y = 37, rule = LifeHistory
26.2A$26.A23.2A$27.3A6.2A12.2A17.2A$29.A6.2A31.A$70.A$47.2A20.2A$47.
2A29.A6.A$76.3A5.A.A$C74.A8.A.A$3C72.2A6.2A.3A$D.C15.2A69.A$2.C15.2A
30.C15.2A15.2A.3A$48.D.C15.2A15.2A.A$48.3C$48.C$52.A$54.A$21.2A65.2A$
22.A32.A32.A$21.A13.2A3.2A15.A28.A.A$21.2A13.A3.A45.2A$33.3A5.3A$33.A
9.A5$54.E$54.E.E$54.3E$56.E20.2A$77.A$66.2A10.A$66.2A2.2A5.2A$70.A.A$
72.A$72.2A!

Edit 2: The modification of Fx77 could be used, for example, to reduce the size of the smallest p117 gun in the gun collection. This case is not very interesting, however, as the gun can be improved further with new conduits:

Code: Select all

x = 176, y = 65, rule = B3/S23
130bob2o$130b2obo2$45b2o14bo15bo9bo49b3o28bo$20b2o24bo14b3o13b3o5b3o
29bo19bo22bo6bobo$21bo13b2o6b3o18bo15bo3bo13b2o17b3o18bo19b3o6bobo$21b
obo11b2o6bo19b2o14b2o3b2o13bo20bo36bo8b2ob3o$7b2o13b2o74bo20b2o11b2o
23b2o13bo$7b2o30b2o24b3o30b2o32b2o32b2ob3o$38bo2bo2b2o19bo88bo11b2obo$
38bobo25bo87bo$b2o38bob2obo111bo$b2o34bobo3b2obo34b3o74b2o$5b2o30bob2o
3b3o6b2o25bo2bo34bobo37b3o5b2o$5b2o30bob2o2bobo7b2o28bo17b2o14bo3bo36b
2o6b2o$41b2obo34bo4bo16b2o14bo4bo31b2ob2o$43bo34bo5b3o30bo4bo32b3o$77b
o6b2o32bo3bo21b2o10bo$2o75b4ob3o33bo2bo23bo$2o76b2o3bo58b3o8b2o14b2obo
$9b3o38b2o90bo11bo14bob2o$8bo2bo8bo3b2o24bo19b2o79b3o$8bo3bo6bobo3bo5b
2o3b2o13bo17bobo11b2o6bo19b2o38bo10b2o$7bo2b3o6b2o3bo7bo3bo13b2o17bo
13b2o6b3o18bo20b2o27b2o$8b4o11bo5b3o5b3o28b2o24bo14b3o21bobo$19b5obo3b
o9bo53b2o14bo25bo$19bo4b2o109b2o$12bobo5b3o$13b2o7bob2o$13bo9bobo$152b
2o$153bo$153bobo$154b2o$20b2o$20bobo$22bo$22b2o$150bobo9bo$150b2obo7b
2o$153b3o5bobo$39b2o109b2o4bo$40bo25bo14b2o53bo9bo3bob5o$40bobo21b3o
14bo24b2o28b3o5b3o5bo11b4o$12b2o27b2o20bo18b3o6b2o13bo17b2o13bo3bo7bo
3b2o6b3o2bo$12b2o10bo38b2o19bo6b2o11bobo17bo13b2o3b2o5bo3bobo6bo3bo$
22b3o79b2o19bo24b2o3bo8bo2bo$3b2obo14bo11bo90b2o38b3o$3bob2o14b2o8b3o
58bo3b2o76b2o$30bo23bo2bo33b3ob4o75b2o$19bo10b2o21bo3bo32b2o6bo$18b3o
32bo4bo30b3o5bo34bo$17b2ob2o31bo4bo14b2o16bo4bo34bob2o$8b2o6b2o36bo3bo
14b2o17bo28b2o7bobo2b2obo30b2o$8b2o5b3o37bobo34bo2bo25b2o6b3o3b2obo30b
2o$16b2o74b3o34bob2o3bobo34b2o$17bo111bob2obo38b2o$21bo87bo25bobo$6bob
2o11bo88bo19b2o2bo2bo$4b3ob2o32b2o32b2o30b3o24b2o30b2o$3bo13b2o23b2o
11b2o20bo74b2o13b2o$4b3ob2o8bo36bo20bo13b2o3b2o14b2o19bo6b2o11bobo$6bo
bo6b3o19bo18b3o17b2o13bo3bo15bo18b3o6b2o13bo$6bobo6bo22bo19bo29b3o5b3o
13b3o14bo24b2o$7bo28b3o49bo9bo15bo14b2o!

[dvgrn]

I've noticed that a few conduit connections could be "improved" slightly. I'm not sure if these are in any way useful, however.

I don't see any downside to the Fx77 modification -- if you save a row, you save a row. So I've changed vfx77sw.rle in the current Hersrch build -- ufx77s.rle didn't seem to need changing, since it's already very optimistic about what might be made to fit by welding...!

The Bx222w2 alteration I think I'm going to leave in the same category as Karel Suhajda's reduction of Fx77+R64:

Code: Select all

#C Karel Suhajda, 4 January 2004: Fx77 + R64 variant,
#C replacing two eaters with a block, at the cost of a slower recovery
x = 48, y = 30, rule = B3/S23
37b2o$17bo20bo$17b3o18b3o$20bo$19b2o2$o$3o$3bo$2b2o11b2o$15b2o6$bo44b
2o$bo44b2o$b3o36b2o$3bo36b2o3$42b2o$35b2o5b2o$35b2o4$14b2o$14b2o!

On first glance, the new Bx222w2 is a fairly straightforward improvement over the old one... but I think it'd be hard to get Hersrch to build clean circuits with that variant.

Might have to define a whole series of new variants of all of the possible following conduits, with the first block-suppressing eater removed from each one (?) I could perfectly well be wrong about that, but in the meantime that Bx222w2 variant will probably stay rather low on my priority list. Might possibly have to do something along the same lines anyway (define standard conduits minus the first eater) to get Hersrch to work properly with some of the other new conduits that we haven't tackled yet...

When a variant like this one makes some particular impossible sequence of conduits into a workable sequence, that sequence is supposed to be commented out in the "forbidden" section of data.ini, with the RLE appended so that there's a record of how to do the customization. That might be a good place to keep a record of this new Bx22w2, for the time being, assuming a suitable conduit sequence can be found. (Anyone care to locate one?)

[dvgrn]

Here's what I've come up with so far for R178 and Lx142 -- no forbidden pairs yet. The "GRX" rating already limits the following conduits to the "dependent" ones anyway, F166 and Lx200 and L161 and L214... but specifically forbidding F216, Fx177, and Lx84 looks like a good idea.

data.ini additions:

Code: Select all

Lx142 Lx 142 38 -15

  # This conduit requires the following dependent conduit's glider
  VLX142 1 262 0 GRX data/vlx142.rle 3 8 data/ulx142.rle 2 8

R178 R 178 46 5

  # This conduit requires the following dependent conduit's glider
  VR178 1 265 0 GRX data/vr178.rle 3 9 data/ur178.rle 2 7

ur178.rle:

Code: Select all

45b2o$44b5o3b2o$45b4o2b4o$41bob6o3b2o$39b10o2b2o$29bo9b15o$29b2o7b17ob
o$14b2o6b3o5b2o6b20o$2b2o9b5o4bo6b5o5b15ob3o$b3ob3o6b4ob3o7b6o3b15o3bo
$3b7o2b11o5b6o2b17o$3b22obob24o$4b48o$3b40ob7o$b42o2b6o$43o2b7o$ob40o
4b6o$obob4ob26o10b6o$o3b4o2b16o3b6o11b5o$3ob4o2b7ob4o8b6o10b6o$7o4b6o
2b2o7b9o9b6o$11b6o11b2o4b4o$12b3o14bo5b4o9b3o$10b7o19b4o8bo$10b2obob2o
20b3o7b3o$11bo3bo22bo!

vr178.rle:

Code: Select all

46b2o$46b2o5b2o$53b2o$33bo$31b3o$25b2o3bo20b2o$2o24bo3b2o19b2o$bo13b2o
6b3o31b2o$bobo11b2o6bo33b2o$2b2o7$2bo28bo$2bo$2b3o$4bo$35bo$29b2o$30bo
18b3o$27b3o19bo$11b2o3b2o9bo11bo8b2o$12bo3bo$9b3o5b3o$9bo9bo!

ulx142.rle [EDIT: oops, fixed copy-and-paste typo as per Sokwe's note below]:

Code: Select all

30bo$29b3o7b3o$28b4o8bo$21bo5b4o9b3o$12bo7b2o4b4o$11b2o7b9o9b6o$11b5o
6b6o10b6o$2b2o9b4o4b6o11b5o$b3ob3o4b15o10b6o$3b7ob23o4b6o$3b32o2b7o$4b
31o2b6o$3b32ob7o$b43o$17obob24o$ob13o5b6o2b17o$obob4ob6o6b6o3b15o3bo$o
3b4o2bob5o4b5o5b15ob3o$3ob4o7b2o5b2o6b20o$7o8bo5b2o7b17obo$21bo9b15o$
31b10o2b2o$33bob6o3b2o$37b4o2b4o$36b5o3b2o$37b2o!

vlx142.rle:

Code: Select all

19bo11bo8b2o$10bo8b3o19bo$10b3o9bo18b3o$13bo7b2o$2o10b2o13bo$bo$bobo$
2b2o$23bo6$2bo$2bo$2b3o44b2o$4bo11b2o31b2o$16bo5b2o19b2o$17b3o2bo20b2o
$19bo3b3o$25bo$45b2o$38b2o5b2o$38b2o!

sanity-check-R178-Lx142-L198-Lx212.bat:

Code: Select all

set x=30000
set u=1500
set t=100
set n=250

set name=new-conduit-test

hersrch -p 997 -o %name%.rle -n %n% -f %name%.log -s -x %x% -u %u% -t %t% -e (i=1060..1069)F(0)[0,0]..B(i)[-107,-163]
pause

I'm out of time for tonight, but should have the new Hersrch posted before too much longer -- check the initial message in this thread for a link to a November 2012 Hersrch ZIP file.

[Sokwe]

ulx142.rle:

You seem to have accidentally copied the ur178 data here.

sanity-check-R178-Lx142-L198-Lx212.bat

I hope you are also going to include Lx138 (posted earlier in this topic).

Here's the data for Bx167:

Code: Select all

Bx167 Bx 167 3 -35

  # Standard variant (following glider required)
  VBX167 1 116 0 GRU data/vbx167.rle 1 6 data/ubx167.rle 2 3

  # This conduit requires the following dependent conduit's glider
  VBX167X 1 167 0 GRX data/vbx167x.rle 1 6 data/ubx167x.rle 2 3

vbx167.rle:

Code: Select all

5bo$5bo$3b3o9b2o3b2o$3bo11b2o3b2o$7bo$9bo2$10bo$12bo3$23b2o$8b2o13b2o$
8b2o2$5b2o$5b2o3$4bo$3bobo$4bo$b3o20b2o$bo22b2o12$o$o22b2o$3o20bo$2bo
10b2o3b2o4b3o$11bo2bo4bo6bo$11b3o4bo$18b2o$11b2obo$11bob2o!

ubx167.rle:

Code: Select all

8b2o2b2o2bo$6bob10o$4bobob11o2bo$4b3ob11ob3o$4bo3b15o$8b12obo$8b4ob7o$
9b4ob7o$10b12o$11b11o$12b9ob3o$13b12o$10b2ob13o$9b17o$10b2ob10obo$13b
10o$13b9o$12b10o$11b11o$10b11o$9b13o$8b15o$7b19o$9b18o$10b17o$9b17o$6b
16o$5b16o$5b16o$4b16o$3b20o$3b20o$4b20o$3b22o$b24o$24o$ob13ob10o$obob
4ob7o2b2o4b2o$o3b4o2b3obo4bo4bo$3ob4o4b4o3b2o$7o4b2o2bo$12b3o!

vbx167x.rle:

Code: Select all

5bo$5bo$3b3o9b2o3b2o$3bo11b2o3b2o5$o2$bo$3bo19b2o$23b2o$4bo3$5b2o$6bo$
3b3o$3bo3$24b2o$24b2o12$o$o22b2o$3o20bo$2bo10b2o3b2o4b3o$11bo2bo4bo6bo
$11b3o4bo$18b2o$11b2obo$11bob2o!

ubx167x.rle:

Code: Select all

8b2o2b2o2bo$6bob10o$4bobob11o2bo$4b3ob11ob3o$4bo3b15o$8b12obo$11bob7o$
14b7o$8b3o3b8o$8b4o2b8o$6b6ob8ob3o$6b19o$6b20o$6b20o$7b16obo$8b15o$6b
5ob10o$6b2o5b9o$11b11o$11b10o$10b12o$9b14o$9b17o$10b17o$10b17o$9b17o$
6b16o$5b16o$5b16o$4b16o$3b20o$3b20o$4b20o$3b22o$b24o$24o$ob13ob10o$obo
b4ob7o2b2o4b2o$o3b4o2b3obo4bo4bo$3ob4o4b4o3b2o$7o4b2o2bo$12b3o!

Pairs data:

Code: Select all

VBX167-VL106P6
VBX167-VR218P4
VBX167-VR218P6
VBX167-VR218P10
VBX167-VRX202
VBX167-VRX202X

VBX167X-VLX84P15
VBX167X-VF216P6
VBX167X-VF216P6X
VBX167X-VFX177P6
VBX167X-VFX177P6X

VL214P8-VBX167
VL214P8-VBX167X

[dvgrn]

You seem to have accidentally copied the ur178 data...

Fixed.

I hope you are also going to include Lx138 (posted earlier in this topic).

Done -- Lx138 and Bx167, thanks for both. Made a separate minimal test batch file for these last two conduits:

sanity-check-Lx138-Bx167.bat:

Code: Select all

set x=20000
set u=1500
set t=100
set n=250

set name=new-conduit-test2

hersrch -p 997 -o %name%.rle -n %n% -f %name%.log -s -x %x% -u %u% -t %t% -e (i=300..310)F(0)[0,0]..R(i)[52,-35]
pause

A Spartan search now turns up the appropriate open-ended conduit for VLX142 + VF166 + VR178 + VF166 ... definitely a good sign, though it'd be a good idea to do some more thorough testing.

sanity-check-Spartan.bat:

Code: Select all

set x=30000
set u=1500
set t=100
set n=25

set name=test-new-Spartan

hersrch -p 997 -o %name%.rle -n %n% -f %name%.log -s -x %x% -u %u% -t %t% -e (i=650..660)F(0)[0,0]..Bx(i)[-19,-107]
pause

I've posted the new archive Hersrch-6Nov2012.zip in the usual place in the Hersrch project folder. Additions so far and plans for the next stage are listed in changelog.txt. Lots more conduits left to add -- I don't really want to think exactly how many!

Any problems you find, please let me know.

[dvgrn]

I don't see any downside to the Fx77 modification -- if you save a row, you save a row.

Turns out there _is_ a problem with adding the new narrower Fx77sw to Hersrch. Every time it crops up in the latest output RLE (6 November 2012 Hersrch), it has to be edited to remove the overlapping first fishhook eater from the following conduit. The overlap used to be harmless, because there was a welded eater in the exact same location.

That's not so good. Unfortunately this case is not amenable to the kludgey trick I used on the R178 and Lx142, where there's a line of dots that shows where the dependent-conduit glider will come from. When an eater shows up in the middle of the line, one of the dots will cause it to self-destruct cleanly.

It wouldn't even help to change "v" files to a multi-state rule like LifeHistory that allows OFF cells to be specified; the following conduit will still drop on top of the Fx77SW. So we either just have to learn to live with editing the candidate tracks to remove miscellaneous eaters, every time an Fx77sw shows up (ugh) ... or we can try adding new versions of all the conduits that have an eater in that location: VF117M, VFX119GM, VFX119P2M, VFX119BM, VFX153M, VFX176M, VL112M, VR218P4M, VR218P6M, VR218P10M, VLX134P15M, VLX134P15XM, VLX212M, VLX212XM.

I suggested this a few days back, and it's actually not hard to do (as these things go). I _think_ it's a good idea, both because it allows us to add the new smaller Bx222W2, and should also make it trivial to add conduits that end with Guam's R126. But I'm wondering how much it will slow Hersrch down.

My next version of Hersrch will be a speed-test build, adding the above fourteen variants with the first eater missing, but making no other changes. If I can get clean new Fx77SW conduits in the output RLE without too much of a slowdown, then it will make sense to use the same variants with Bx222W2 and Bx260 and any other conduits with this new following-block dependency.

[dvgrn]

My next version of Hersrch will be a speed-test build, adding the above fourteen variants with the first eater missing, but making no other changes.

Fifteen variants, actually -- there was a very old data bug in the VFX77SW pairs section. VFX77SW+VFX176 was allowed, but VFX77SW+VFX176X wasn't, so Hersrch for years has probably not finding occasional valid tracks containing VFX77SW+VFX176+[dependent conduit]... I'll run some tests to make sure I'm understanding all this correctly. Anyway, so we now have a VFX176XM as well as a VFX176M.

Oddly enough, everything seems to be working very nicely so far!

Running test.bat gives a matrix of all valid conduit pairs. Hersrch crashes mysteriously on my Windows 7 machine after producing the matrix, but the matrix is the useful part.

Over the next few days I'll be carefully reviewing all the conduits I've added, running some more searches and doing a few real speed tests. However, it seems like a good idea to release a beta version of Hersrch at this point, with the new first-eater-missing 'm' conduits:

http://cranemtn.com/life/files/Hersrch/ ... ov2012.zip

If anyone sees any problems with search output, let me know as usual -- thanks! This is more of a beta release than usual, so I haven't updated the official link at the top of this thread yet. I'll do that after adding the next two Spartan conduits. Meanwhile, it would be great if someone could independently verify that I haven't slowed things down too much from the November 6 version.

My first attempt to add the 'm' conduits ran into trouble because most of the variants with the eater missing were slightly smaller than any other variants, so Hersrch chose them first whenever possible -- at the very beginning of a candidate track, especially. I ended up dodging that problem by adding a dot five cells behind the input Herschel, so that the 'm' variants would be slightly larger than the standard ones. This is a fairly kludgey workaround, and it may not work in certain cases -- but it allows the project to move forward, so I'm happy with it for now. Here's some sample output with the marker dots on the 'm' input Herschels:

Code: Select all

#C VFX77SW VFX119GM VFX77
x = 74, y = 43, rule = B3/S23
45bo$45b3o$48bo$47b2o11b2o11bo$60b2o9b3o$27bo43bo$27b3o5b2o34bo$30bo4b
2o$29b2o2$46bo$30b2o14bo$30b2o3b2o9b3o$35b2o11bo$44bo$42bo2$41bo$39bo$
61b2o$61bobo$63bo$o62b2o$3o$3bo43b2o$2b2o11b2o11bo18b2o$15b2o6bo2b3o$
26bo$26bo2$50bo$48b3o$bo45bo$bobo17b2ob2o20bobo$b3o18bobo22bo$3bo18bob
o$21b2obob2o$18b2o2bobobobo16b2o$18b2obo2bo2b2o16b2o$21bob2o16b2o$18b
2obobo17bo$12b2o4bo2bobo18b3o$12b2o6b2ob2o19bo!

Code: Select all

#C VBX222W2 VF117M
x = 82, y = 41, rule = B3/S23
21b2o$21bo$22bo11bo$21b2o9b3o$31bo16b2o$31b2o15bo$46bobo12b2o$46b2o13b
o$62bo$2bo15b2o19b2o20b2o$2bo15b2o19b2o29bo6bo$3o65b3o5bobo$o66bo8bobo
$67b2o6b2ob3o$81bo$42bo15b2o15b2ob3o$42bo15b2o15b2obo$40b3o2bo$27b2o
11bo$28bo15bo$25b3o18bo$25bo54b2o$47bo32bo$49bo28bobo$78b2o7$46bo$46bo
bo$46b3o$48bo20b2o$69bo$58b2o10bo$58b2o2b2o5b2o$62bobo$64bo$64b2o!

Karel Suhajda recently pointed out an error in my registration of Brice Due's F171p15, so that's another very minor long-standing inaccuracy in search results that has now been fixed. I have also included Sokwe's improved Bx222w2, as shown above.

[Sokwe]

When looking through the pairs data, I noticed that there were relatively few forbidden pairs that started with VL106P6, despite the fact that it can only be followed by a few conduits. I suggest removing pairs data that starts with "VL106P6-" and adding the following:

Code: Select all

VL106P6*
VL106P6+VBX167
VL106P6+VBX167X
VL106P6+VL198
VL106P6+VLX65P3B
VL106P6+VLX65P4B
VL106P6+VLX65P5B
VL106P6+VLX65P5B1
VL106P6+VLX65P7B
VL106P6+VLX65P8B
VL106P6+VLX65P22B
VL106P6+VLX73P5
VL106P6+VLX73P6
VL106P6+VLX73P7
VL106P6+VLX73P8A
VL106P6+VLX73P9
VL106P6+VLX73P22

Bx167 and L198 can follow L106p6 if the snake stabilizing the bookend is replaced with a skewed bookend. The Lx73 conduits can follow L106p6 if the unix near the output Herschel is replaced with a pipsquirter like so:

Code: Select all

x = 44, y = 34, rule = B3/S23
5bo2bo$8bo$4bo4bo$4b2o2bobo$9bob2o11b3o$24bo$23b3o$9bo2bo$9b2o4$4b2o$
2o2b2o$2o2bob2o26bo$5b3o26bo$34b3o$36bo$5b2o$5b2o$38bo$38bo2$26b2o8bo
3bo$27bo6b3obob3o$24b3o6bo4b2o3bo$24bo7bobob2o2b2obo$32bob2o2b2o2bo$
31b2obob2o2b2o$29bo2bobobo2bo$29b2o2b2ob4o$35bo$35bobo$36b2o!

I have also included Sokwe's improved Bx222w2, as shown above.

Is there a reason you used the variant with the eater rather than this one with a block?

Code: Select all

x = 82, y = 41, rule = B3/S23
21b2o$21bo$22bo11bo$21b2o9b3o$31bo$31b2o9b2o$42b2o17b2o$61bo$62bo$2bo
15b2o19b2o20b2o$2bo15b2o19b2o29bo6bo$3o65b3o5bobo$o66bo8bobo$67b2o6b2o
b3o$81bo$42bo15b2o15b2ob3o$42bo15b2o15b2obo$40b3o2bo$27b2o11bo$28bo15b
o$25b3o18bo$25bo54b2o$47bo32bo$49bo28bobo$78b2o7$46bo$46bobo$46b3o$48b
o20b2o$69bo$58b2o10bo$58b2o2b2o5b2o$62bobo$64bo$64b2o!

[dvgrn]

I have also included Sokwe's improved Bx222w2, as shown above.

Is there a reason you used the variant with the eater rather than this one with a block?

Nope, just a mistake on my part -- I knew about the block version last week, but this morning I forgot and used the eater variant. I'm fixing it now in my sandbox, so the block version will show up in the next build.

Thanks for the notes on VL106P6 -- I'm going to roll those changes in right away before I forget. I suspect there's quite a lot more forbidden-pair optimization left to do. I know I've added a couple of conduits without taking the time to deal with forbidden pairs at all. Now that I've noticed that Hersrch's test.bat test-matrix function does work for me after all, it will be much easier to do that part of the registration.

[Kasuha]

When looking through the pairs data, I noticed that there were relatively few forbidden pairs that started with VL106P6, despite the fact that it can only be followed by a few conduits. I suggest removing pairs data that starts with "VL106P6-" and adding the following:

Hersrch always tests compatibility of pairs by overlapping their usage diagrams, the pairs section is only there to override this default behavior. I have checked the test pattern (which was added exactly for this purpose) and it seems to me that almost all of what's left is compatible, either as is or with the pipsquirter. So I suggest against using fully manual definition because you'll likely forget to add pairs with components added later - rather just solve the differences.

[Sokwe]

Hersrch always tests compatibility of pairs by overlapping their usage diagrams, the pairs section is only there to override this default behavior. I have checked the test pattern (which was added exactly for this purpose) and it seems to me that almost all of what's left is compatible, either as is or with the pipsquirter. So I suggest against using fully manual definition because you'll likely forget to add pairs with components added later - rather just solve the differences.

I see why it might be bad to add "VL106P6*", as one would need to remember this when adding future conduits that might be compatible. I disagree with you when you say "almost all of what's left is compatible." L106p6 runs into the Fx77-type eater and the F117-type eater even though their usage diagrams do not overlap, and this seems to be the majority of conduits. The only compatible conduits seem to be Bx167, L198, and variants of Lx65 and Lx73.

[Kasuha]

I disagree with you when you say "almost all of what's left is compatible."

You're right, I was a bit too optimistic about it.

[Sokwe]

As Karel pointed out, adding "VL106P6*" means that anyone who adds a new conduit that is compatible with it would need to know this to correct the pairs data. That is, if they failed to allow the pair, then Hersrch would fail to find some conduits that should work. For this reason, I suggest removing what I gave earlier and replacing it with this:

Code: Select all

# VL106P6-VBX167: 38b2o$22b2o14bo$11b2o9b2o12bobo$11b2o23b2o2$2b2o$2b2o$38b2obo$38bob2o2$2b2o39b2o$2b2o38bobo$38b2o2bo$38bobob2o$8bo31bo$39b2o4bo2bo$7bo37bo$5bo6b2o30bo4bo$12b2o29bobo2b2o$4bo36b2obo$15b2o$3o12b2o3bo$2bo16bobo19bo2bo$2b2o16bobo20b2o$22bo14b2o$22b2o13bo$35b3o7$44b2o$25bo18bobo$25bo20bo$25b3o18b2o$27bo11$25bo2bo$25bo$24bo4bo$23bobo2b2o$21b2obo$35b2o$35b2o2b2o$21bo2bo8b2obo2b2o$23b2o8b3o3$34b2o$34b2o!
VL106P6-VBX125
VL106P6-VBX222
VL106P6-VBX222W1
VL106P6-VBX222W2
VL106P6-VF117
VL106P6-VF171
VL106P6-VF349P15
VL106P6-VF349P15X
VL106P6-VFX77
VL106P6-VFX78P8
VL106P6-VFX158
VL106P6-VFX176
VL106P6-VFX349P15
VL106P6-VFX349P15X
VL106P6-VL112
VL106P6-VL156
# VL106P6-VL198: 59b2o$43b2o14bo$43b2o12bobo$57b2o4$59b2obo$59bob2o$o$3o61b2o$2bo15b2o43bobo$2bo15b2o39b2o2bo$59bobob2o$61bo$30bo29b2o4bo2bo$29bobo34bo$30b2o33bo4bo$21b2o41bobo2b2o$22bo39b2obo$21bo$21b2o$62bo2bo$64b2o$58b2o$58bo$56b3o7$65b2o$46bo18bobo$46bobo18bo$46b3o18b2o$48bo11$46bo2bo$46bo$45bo4bo$44bobo2b2o$42b2obo$56b2o$56b2o2b2o$42bo2bo8b2obo2b2o$44b2o8b3o3$55b2o$55b2o!
VL106P6-VLX65P3
VL106P6-VLX65P3X1
VL106P6-VLX65P3X2
VL106P6-VLX65P3
VL106P6-VLX65P3X1
VL106P6-VLX65P3X2
VL106P6-VLX65P8
VL106P6-VLX65P8X1
VL106P6-VLX65P8X2
VL106P6-VLX65P4
VL106P6-VLX65P4X1
VL106P6-VLX65P4X2
VL106P6-VLX65P5
VL106P6-VLX65P5X1
VL106P6-VLX65P5X2
VL106P6-VLX65P7
VL106P6-VLX65P7X1
VL106P6-VLX65P7X2
# VL106P6-VLX73: 25b2o$24bo2bo$24bobobo$23b2obobo$25bobob2o$20b2o2b2obo2bo$26bobobob2o$16b2o5b2obob3o2bo$16bo7bo2bo3b2o$14b3o7bob5o$25bo4bo$26b4o$28bo$30bo$29b2o2$23b2o$4bo18bobo$4bobo18bo$4b3o18b2o$6bo11$4bo2bo$4bo$3bo4bo$2bobo2b2o$2obo$14b2o$14b2o2b2o$o2bo8b2obo2b2o$2b2o8b3o3$13b2o$13b2o!
VL106P6-VLX138
VL106P6-VLX142
VL106P6-VLX212
VL106P6-VLX318P15
VL106P6-VR190 
VL106P6-VR190X
VL106P6-VR318P15

[Edited to include all previous suggestions]

@Dave
In vbx222w2.rle, you seem to have left the input Herschel intact instead of using a ghost Herschel. Also, please add this to the pairs data:

Code: Select all

VL214P8-VBX167
VL214P8-VBX167X

[dvgrn]

In vbx222w2.rle, you seem to have left the input Herschel intact instead of using a ghost Herschel. Also, please add this to the pairs data:

Code: Select all

VL214P8-VBX167
VL214P8-VBX167X

Done and done.

I assume the forbidden L106p6 pairs you quoted are in addition to the pairs that I recently deleted in favor of VL106P6*... You don't have any new magic tricks for VBX125, VBX222, VFX158, VL112, or VR190, right?

I've gotten the rest of the Spartan conduits registered now -- Rx140(x), R126, Bx198, and Bx260, along with a miscellaneous non-Spartan F114p8(b) for good measure. So there will be a new Hersrch archive out as soon as I can find an hour or so to double-check all the new additions.

[dvgrn]

... there will be a new Hersrch archive out as soon as I can find an hour or so to double-check all the new additions.

I've run a fair number of trial searches, and the new conduits seem to be showing up and working as expected. A new ZIP archive is in the usual place (13 November 2012). Conduits added since the last beta are F114p8(b), Rx140(x), R126, Bx198, and Bx260 -- the last three are Spartan, as is the Rx140x variant. Let me know if you run into any problems.

I haven't done anything yet with forbidden pairs or sequences for the conduits I've added -- except for Bx260, which I've done the easy but sub-optimal way for now:

Code: Select all

VBX260*
VBX260+VF171
VBX260+VR190
VBX260+VR190X

I'll gradually work through the test.bat matrix with an eye to adding more forbidden pairs to the next release.

[Sokwe]

I assume the forbidden L106p6 pairs you quoted are in addition to the pairs that I recently deleted in favor of VL106P6*... You don't have any new magic tricks for VBX125, VBX222, VFX158, VL112, or VR190, right?

Ah, you're right... oops. I was adding them based on the test pattern from an earlier version, so I missed those.