RuffledBricks

Speedrunning seems to be a fundamentally different cultural way to play JSW, involving a different mindset, compared with what I'm accustomed to!  But getting into the spirit, would it not be an interesting tactic to deliberately sacrifice all of Willy's lives (bar his last one) at the very start of the game?  So that the game doesn't have to process drawing any remaining lives on the status bar from that point onwards.  (Of course, you'd need to be sure not to get killed again after that point!)
Thinking about Manic Miner, different aspects may be in play.  Do you still go by the Real World timings, given that the final score is a more precise reflection of how quickly you completed all the caverns (based on air supply which saps in more discrete doses, compared with the minutes occasionally ticking over in JSW)?
Actually, I've just thought of something else - do you deliberately NOT collect the bonus points for making King fall off his perch when you're speedrunning MM?  Not just because it could be considered an unnecessary diversion if reaching the Final Barrier as quickly as possible is your only concern, but also because collecting the bonus points for dislodging Kong actually brings forward the point where you win a bonus life - which has the inadvertent consequence of slowing down the game sooner (due to the program having to draw an extra life earlier)?

The colours are from an lcd widescreen monitor. Nothing special.~

Provided here is the final version of the random patch. This is similar to the listing I provided, but is extended in scope.
Using an original JSW file - which I am not providing.
 
 
; Load JSW  - ORIGINAL  - RETURN TO BASIC
; load this file over JSW - a 238 byte file - code loads at 65400
; RANDOMIZE USR 65400  - TO INSTALL PATCH
; RANDOMIZE USR 33792  - TO PLAY

Not extensively tested - main purpose is to see what the game runs like when modified in such a way.
 
This file is one byte bigger. Which shifts the room data by one byte and corrects the problem (I hope)
 
 
 
rand_sp.tap

The technical changes from the original file I listed.

It was noticed that out of all the playable rooms. Only 6 needed data to change how the routine modified the room. So this file stores only six room numbers and the data for those 6 rooms ( and an extra one for luck- CPIR)

The original query wanted a period of immunity after death. Whilst easy enough to do. It needs more code than I was prepared to write. The code is not the problem. The backward reference to the original is.

After playing the modified rooms I noticed I was occasionally killed as I entered a room. The random aspect meant that I would re-enter the room and again in most cases the sprite would have been moved by the routine, and would start in another place. So respawning back in the same place was not a fatal flaw.

After wandering around heaps of rooms, it became noticeable that this occasional problem was mainly the horizontal sprites. (it might happen with vertical sprites - but not on my wandering) So instead of addressing the immunity problem I instead addressed the problem of sprites being in the vicinity of room edges instead.

The added code that is not in the original listing, takes the trouble of moving any sprite that is left near a room edge. This added code will move sprites away from the immediate vicinity of the left and right edge. It only moves horizontal sprites.

From the wanderings I had, this small change allows in most cases the entry into a room without a problem.   

On second thoughts: 
If my RZX-recording hosted on RZX Archive shows JSW completed at 7:56 am in-game time and you completed the game with an in-game time of 7:57 am, certainly there is some room for improvement. And it might very well allow you to go sub-20.
Admittedly, you completed the game just a couple of seconds after the in-game clock turned to 7:57, so you were very close to 7:56. However, I created that recording at a time when my personal (Rollback-assisted) completion time of JSW without any unnecessary loss of life was 7:59 am. I've brought it down to 7:58 since, but I have not tried to improve my with-lives-lost 7:56 am recording. So I would not rule out that repeating the route taken and movements executed in my 7:56 with-lives-lost recording, but with the optimisations I must have applied in my later no-lives-lost 7:58 recording, one could go as low as 7:55 with lives lost. Plus there may be some optimisations that you apply that I have missed (like the On The Roof ceiling warp, although I am not 100% sure that it really saves time).
That would mean to me that you could *certainly* go sub-20 on your time scale provided you executed perfectly, while playing live, all of the movements that I was able to execute thanks to dozens (or even hundreds) of Rollback instances.
So there's a challenge for you... 🤪

RuffledBricks, I would like to spend more time following your "work" on JSW and watching your videos, but time is sparse, unfortunately.
I meant to comment on it more in detail, "properly" (according to my standards), but I can't do it right now. So I'll just comment as much - or as little - as and when possible.
Your speedruns are most impressive, taking into account that you are doing it without the possibility of repeating your movements in case of problems 👍.
For years now I have been RZX-recording efficient walkthroughs of MM and JSW games made using Rollback. Most of them are hosted on RZX Archive, and some that are not hosted on RZX Archive (because someone else has submitted their recording first) can be found on my website JSW Central.
Of course, making RZX recordings using the Rollback feature is a lot easier than playing the game live, without the possibility of correcting your errors. Infinitely easier, I believe.
As for the original "JSW", my personal best when using Rollback is finishing the game at 7:58 am in-game time without losing any lives, and at 7:56 am in-game time when losing lives in order to speed up the completion process.
I always use in-game time to determine completion time, so your timing - real-world minutes - seems like temperature expressed in degrees Fahrenheit to me 😄. However, when I now look at my YouTube video of my completion of "JSW" without losing lives, I can see that I started the game at 1:14 into the video and reached the toilet at 24:25. That would make my completion time (Rollback-assisted, of course) 23:11. And that's without any unnecessary loss of life, so it can still be improved by losing lives. However, I am not sure how that time compares to your time - taking into consideration the frame rate of the video, etc. It can't be the same time, I believe, since your in-game completion time at 20:24 was 7:57 am, and my in-game completion time at 23:11 was 7:58 am. Evidently, our real world clocks were not running at the same speed 😀.
You lost 12 second in the Forgotten Abbey, so your completion time could have been  20:12. I think that going under 20 minutes might be possible, but I wouldn't necessarily bet on it.
I did not see any major possibilities for improvement in your video, apart from the beginning and end - is it really faster to collect the "Top Landing" item last? I always do it right after the Bathroom, it seems quicker to me, but I've never tried timing it.
You can watch my video to see if I did anything more efficiently than you did. Also, if you have a look at my recording hosted on RZX Archive (it should be there within the ZIP archive), it's the one where I lose lives to speed up the completion. You could have a look to see if I lose lives in the same places as you do - perhaps if I chose different rooms to lose lives (I don't recall off the top of my head what I did, it's an old recording), it might give you some ideas for improvement 🙂.

Today I managed to get the Any% world record down to 20:24:

https://www.twitch.tv/videos/931996825
There are several new timesaves in this, mostly found by another runner called Mick_jo, with a few from myself as well. These include:
Doing Main Stairway without nipping into Ballroom West to reset the cycle Running under the Jelly in Tree Root on the first cycle Slightly faster execution of the Out On A Limb shortcut by doing a vertical jump over the Bird Smooth stair clip at the bottom of Foot Of Megatree Entirely wait-free execution of East Wall Base On The Roof ceiling warp (actually quite an old strat now, but I hadn't discovered this before my previous Any% run) Timing trick in Ballroom East to ensure you can run under the three Skulls without pausing New jump timing in Ballroom West to avoid having to return through the table to collect the last item Sadly, I had an enormous brainfart in Forgotten Abbey due to a new strat not being quite bedded in properly, so I lost a fair chunk of the saved time on that. Will try it again next week to try and shave that off.
I'm intrigued to know if sub-20 mins is possible, so if anyone's spotted anything I've missed optimisation-wise, please do let me know.
 
 

Has anyone other than @jetsetdanny shared videos of how they got these scores? Curious to know if there's any strats I've been missing from my full game speedruns.

DSCF9962.MOV The centipede moving with a smoother path.    

Today I managed to get the Any% world record down to 20:24:

https://www.twitch.tv/videos/931996825
There are several new timesaves in this, mostly found by another runner called Mick_jo, with a few from myself as well. These include:
Doing Main Stairway without nipping into Ballroom West to reset the cycle Running under the Jelly in Tree Root on the first cycle Slightly faster execution of the Out On A Limb shortcut by doing a vertical jump over the Bird Smooth stair clip at the bottom of Foot Of Megatree Entirely wait-free execution of East Wall Base On The Roof ceiling warp (actually quite an old strat now, but I hadn't discovered this before my previous Any% run) Timing trick in Ballroom East to ensure you can run under the three Skulls without pausing New jump timing in Ballroom West to avoid having to return through the table to collect the last item Sadly, I had an enormous brainfart in Forgotten Abbey due to a new strat not being quite bedded in properly, so I lost a fair chunk of the saved time on that. Will try it again next week to try and shave that off.
I'm intrigued to know if sub-20 mins is possible, so if anyone's spotted anything I've missed optimisation-wise, please do let me know.
 
 

Today I managed to get the Any% world record down to 20:24:

https://www.twitch.tv/videos/931996825
There are several new timesaves in this, mostly found by another runner called Mick_jo, with a few from myself as well. These include:
Doing Main Stairway without nipping into Ballroom West to reset the cycle Running under the Jelly in Tree Root on the first cycle Slightly faster execution of the Out On A Limb shortcut by doing a vertical jump over the Bird Smooth stair clip at the bottom of Foot Of Megatree Entirely wait-free execution of East Wall Base On The Roof ceiling warp (actually quite an old strat now, but I hadn't discovered this before my previous Any% run) Timing trick in Ballroom East to ensure you can run under the three Skulls without pausing New jump timing in Ballroom West to avoid having to return through the table to collect the last item Sadly, I had an enormous brainfart in Forgotten Abbey due to a new strat not being quite bedded in properly, so I lost a fair chunk of the saved time on that. Will try it again next week to try and shave that off.
I'm intrigued to know if sub-20 mins is possible, so if anyone's spotted anything I've missed optimisation-wise, please do let me know.
 
 

I just imagined a randomised version of Attic Centipede and laughed out loud.

This has been tested and works exactly as specified. It needs the room data changing to stop the giant heads being shifted and the forgotten Abbey and the attic data being shifted- An easy enough task
Half an hour to write.
Somewhere near the end of the room set up a call is made to rephase_room
The data at offset #e0 in every room is changed to #00 modify the sprites or #ff leave alone. Or the bits in the byte signal which sprites must be left alone. So writing 1110 0000b would instruct the routine to leave the first three sprites alone. ( eg. stop it phase shifting the big heads if they are the first 3 sprites in a room)
 
 

random_sp      equ    #80e0
return_2_me    equ    #91b6
entity1             equ    #8100

; rephase a room
rephase_room:
    ld    hl,random_sp     ; this address is any free byte in the room data - each room needs this setting up - most rooms can be left as zero
    ld    a,(hl)
    cp #ff
    ret z
    ld    c,a        ; this is the entity bit move mask
;initialise the guardians
    ld    a,#c9
    ld    (return_2_me),a   ; modify the next sprite entry  - return2 me is at address #91b6
    LD IX,entity1        ;  Point IX at the first byte of the first entity buffer at L8100
Phase_shift:
;test the end of sprite data then test the bit held in c; this signals a possible move
    LD    A,(Ix+0)
    inc a  ; cp #ff
    jr    z,no_more
    bit    7,c
    jr    nz,a_static
    cALL    RANDOM
    AND    127
    INC    A ; instead of INC B 
    LD    B,A
SHIFT_IT
    PUSH    BC
    CALL    L90C4        ; MOVE THE SPRITE
    POP    BC
    DJNZ    SHIFT_IT
a_static
    ld    de,8
    add    ix,de
    rlc    c
    jr    Phase_shift
; ensure it is put back
no_more
    ld    a,#11            ;ld de,#
    ld    (return_2_me),a
  ret
 
; the RANDOM routine does not produce RANDOM numbers it produces a sequence of numbers that is fixed in sequence. The number of digits in the sequence is large before it repeats. At machine speed calling random will produce a pattern based on its output.
RANDOM:
    PUSH    HL
S_M_C_address equ    $+1
    LD    HL,$-$
S_M_C_SEED     equ    $+1
    LD    A,$-$
    xor    (HL)
    XOR    L
    XOR    H
    inc    hl
    res    5,h
    LD    (S_M_C_address),HL
    LD    (S_M_C_SEED),A
    POP    HL
    RET
; *** NOTE ***  only 6 rooms have a value that is not 0 (zero) - 7,20,28,39,42,48 ---- rooms value start at 1 in this list
; values for byte #e0
;"The Off Licence" 1
 db #00
;"The Bridge" 2
 db #00
;"Under the MegaTree" 3
 db #00
;"At the Foot of the MegaTree" 4
 db #00
;"The Drive" 5
 db #00
;"The Security Guard" 6
 db #00
;"Entrance to Hades" 7
 db 11100000b    ;<<<<<<<<<<<<<<
;"Cuckoo's Nest" 8
 db #00
;"Inside the MegaTrunk" 9
 db #00
;"On a Branch Over the Drive" 10
 db #00
;"The Front Door" 11
 db #00
;"The Hall" 12
 db #00
;"Tree Top" 13
 db #00
;"Out on a limb" 14
 db #00
;"Rescue Esmerelda" 15
 db #00
;"I'm sure I've seen this before.." 16
 db #00
;"We must perform a Quirkafleeg" 17
 db #00
;"Up on the Battlements" 18
 db #00
;"On the Roof" 19
 db #00
;"The Forgotten Abbey"  20
 db #ff      ;<<<<<<<<<<<<<<<
;"Ballroom East" 21
 db #00
;"Ballroom West" 22
 db #00
"To the Kitchens Main Stairway"  23
 db #00
;"The Kitchen" 24
 db #00
;"West of Kitchen" 25
 db #00
;"Cold Store" 26
 db #00
;"East Wall Base" 27
 db #00
;"The Chapel" 28
 db 11100000b       ;<<<<<<<<<<<<<<<
;"First Landing" 29
 db #00
;"The Nightmare Room"  30
 db #00
;"The Banyan Tree" 31
 db #00
;"Swimming Pool" 32
 db #00
;"Halfway up the East Wall" 33
 db #00
;"The Bathroom" 34
 db #00
;"Top Landing"  35
 db #00
;"Master Bedroom" 36
 db #00
;"A bit of tree" 37
 db #00
;"Orangery" 38
 db #00
;"Priests' Hole" 39
 db 11100000b       ;<<<<<<<<<<<<<<<
;"Emergency Generator"  40
 db #00
 db Jones will never believe this"  41
 db #00
;"The Attic" 42
 db #ff      ;<<<<<<<<<<<<<<<
;"Under the Roof" 43
 db #00
;"Conservatory Roof" 44
 db #00
;"On top of the house" 45
 db #00
;"Under the Drive"  46
 db #00
;"Tree Root"  47
 db #00
;"[" 48
 db #FF    ; no sprite data       ;<<<<<<<<<<<<<<<
;"Nomen Luni" 49
 db #00
;"The Wine Cellar" 50
 db #00
;"Watch Tower" 51
 db #00
;"Tool Shed" 52
 db #00
"Back Stairway" 53
 db #00
;"Back Door" 54
 db #00
;"West Wing"  55
 db #00
;"West Bedroom" 56
 db #00
;"West Wing Roof" 57
 db #00
;"Above the west Bedroom"  58
 db #00
;"The Beach" 59
 db #00
;"The Yacht"  60
 db #00
;"The Bow"  61
 db #00

The data followed by ;<<<<<<<<<<< is indicating values that are NOT zero
 
 
 
 
 

Here's an idea I had for a mod, though I do not have the programming know-how to create this myself.
I'd be interested to play a version of JSW that was exactly the same as the original (with the pokes, natch), except all the enemy cycles are randomised, so that each enemy starts from a random position in their cycle when you enter a room.
The pros of this idea are that it would create an extra challenge in the game, as compared to the vanilla game you'd never know where the enemies were going to start from in each room and would have to adjust your timings and strats on the fly as a result.
A possible downside of this idea is that this could potentially lead to more instant death loops if the randomiser places an enemy right at the entrance of a room - for example, you could die immediately in Top Landing if the Swiss army knife enemy started their cycle right at the Bathroom door. Although this could also be seen as an upside, as it would require the player to be extra vigilant about the route they take around the mansion to avoid as many instant death 'hot spots' as possible.
Anyhow, what do you all think? Has it been done before? And if it hasn't, how feasible would it be to create?
 
 
 

This is a good point. In my head I hadn't really settled on whether the randomiser would apply once per game (i.e. the positions are randomly set before the game starts and are then fixed once you start) or whether the randomiser would apply every time you enter a room. I hadn't considered that the latter would help to avoid the IDS.
 

(For the unfamiliar, IDS = Infinite Death Scenario.)
RB's idea wouldn't necessarily set up an IDS if Willy entered a room at a point where a guardian started its path, because after losing a life and the room being refreshed, the random start position of that guardian would most likely be somewhere else (whilst Willy's start position in the room would remain the same).
One possible scenario you might get in, say, Forgotten Abbey, could be two monks in sync directly above/below each other, and remaining that way (assuming their horizontal trajectories are of the same width), making it impossible to jump over the lower one without Willy's head hitting the upper one's legs!  But if you walked out of the room and back again, they would reset to different start positions so the problem would be resolved.

I like that idea in principle.
In theory the 'start position' could be likely placed to avoid M.W initial contact, perhaps say a choice of starting the guardian at any number of fixed blocks along its path, avoiding starting at the screen edges and/or near another entry point if MW can safely drop into / climb up into the screen too.
The alternative 'fix' would be to give him a few seconds immunity after a life loss, to allow him to get himself out of the situation but that still means its cost a life to get this reward. Modded JSW2 offers this and makes a reasonable attempt at preventing an instant death situation by restarting the last position but this can actually cause an IDS to happen sometimes! (there are two 'base modded' versions of this 'few seconds immunity' version) one that fits in to the 48K machine with just that extra and one with cheats and a teleport that needs a 128K machine, merely because there was no sane space to fit it into the 48K, but this is off topic really as the JSW2 'engine' is not quite the same as the JSW one, the latter being much easier to 'play with'
Back on topic, I do like the idea provided the starting positions are set within block limits. Some screens care would have to be taken to avoid guardian contact with each other, Forgotten Abbey would be a prime candidate for this as they do overlap each other quite a bit.
It does not sound that difficult to code in theory however it would not really be for me to decide upon that, I have a vague idea on how to set / alter the positions at random but it would likely be a mess!

Here's an idea I had for a mod, though I do not have the programming know-how to create this myself.
I'd be interested to play a version of JSW that was exactly the same as the original (with the pokes, natch), except all the enemy cycles are randomised, so that each enemy starts from a random position in their cycle when you enter a room.
The pros of this idea are that it would create an extra challenge in the game, as compared to the vanilla game you'd never know where the enemies were going to start from in each room and would have to adjust your timings and strats on the fly as a result.
A possible downside of this idea is that this could potentially lead to more instant death loops if the randomiser places an enemy right at the entrance of a room - for example, you could die immediately in Top Landing if the Swiss army knife enemy started their cycle right at the Bathroom door. Although this could also be seen as an upside, as it would require the player to be extra vigilant about the route they take around the mansion to avoid as many instant death 'hot spots' as possible.
Anyhow, what do you all think? Has it been done before? And if it hasn't, how feasible would it be to create?
 
 
 

Nah. So it ended up being 3 and a half minutes on this occasion, although that's largely because I didn't have the tech restart the timer for the first three failed attempts. But the fourth (successful) attempt was a bit sluggish as at that point I was just trying to get it right, so I doubt if I timed that one individually it would be a personal best. For speedrun marathons, you're focusing more just on completing the run than you are on getting a PB time, just because it's done in front of a big audience and you need to give them a satisfactory ending.
I'm thinking of trying to optimise my Writetyper route on stream at some time to see if I can get it even quicker (sub-2 mins would be awesome if I can figure that out), but I probably won't get around to that for a while.

Nah. So it ended up being 3 and a half minutes on this occasion, although that's largely because I didn't have the tech restart the timer for the first three failed attempts. But the fourth (successful) attempt was a bit sluggish as at that point I was just trying to get it right, so I doubt if I timed that one individually it would be a personal best. For speedrun marathons, you're focusing more just on completing the run than you are on getting a PB time, just because it's done in front of a big audience and you need to give them a satisfactory ending.
I'm thinking of trying to optimise my Writetyper route on stream at some time to see if I can get it even quicker (sub-2 mins would be awesome if I can figure that out), but I probably won't get around to that for a while.

Nah. So it ended up being 3 and a half minutes on this occasion, although that's largely because I didn't have the tech restart the timer for the first three failed attempts. But the fourth (successful) attempt was a bit sluggish as at that point I was just trying to get it right, so I doubt if I timed that one individually it would be a personal best. For speedrun marathons, you're focusing more just on completing the run than you are on getting a PB time, just because it's done in front of a big audience and you need to give them a satisfactory ending.
I'm thinking of trying to optimise my Writetyper route on stream at some time to see if I can get it even quicker (sub-2 mins would be awesome if I can figure that out), but I probably won't get around to that for a while.

So here's the ESA VOD for anyone who didn't manage to catch it live:
https://www.twitch.tv/videos/912384624
The JSW Writetyper run met the donation incentive, so it went ahead (timestamp: 22:44:40). As you'll see, nerves got the better of me a little bit - there were 5,000+ people watching so go figure! - and it took four attempts to complete the run on this occasion, but I managed it in the end. It's at least worth watching for the chat comments, which are 90% "what the hell's going on" and 10% "he looks like Mark Ruffalo". 

All this twitch stuff is too complicated for me, but well done anyway, in any money you raised for a good
cause. 

So here's the ESA VOD for anyone who didn't manage to catch it live:
https://www.twitch.tv/videos/912384624
The JSW Writetyper run met the donation incentive, so it went ahead (timestamp: 22:44:40). As you'll see, nerves got the better of me a little bit - there were 5,000+ people watching so go figure! - and it took four attempts to complete the run on this occasion, but I managed it in the end. It's at least worth watching for the chat comments, which are 90% "what the hell's going on" and 10% "he looks like Mark Ruffalo". 

One more Miner Willy-related marathon thing I'd like to plug if I may. 
This Sunday at 10:45am, I'm doing European Speedrunner Assembly's Winter Marathon 2021 in aid of the Swedish Alzheimer's Foundation. I'm only officially booked to speedrun Yazzie, BUT...
They're going to be running an incentive so that if enough people donate to the cause I'll be doing my sub-3 minute Jet Set Willy WRITETYPER speedrun with commentary straight after. So I may end up having to do my silly feat of ridiculously precise keyboard mashing live, rather than in the comfort of my own streams in which I can just do numerous takes until I get it right.
Twitch channel is here if anyone's interested:
https://www.twitch.tv/esamarathon
 

One more Miner Willy-related marathon thing I'd like to plug if I may. 
This Sunday at 10:45am, I'm doing European Speedrunner Assembly's Winter Marathon 2021 in aid of the Swedish Alzheimer's Foundation. I'm only officially booked to speedrun Yazzie, BUT...
They're going to be running an incentive so that if enough people donate to the cause I'll be doing my sub-3 minute Jet Set Willy WRITETYPER speedrun with commentary straight after. So I may end up having to do my silly feat of ridiculously precise keyboard mashing live, rather than in the comfort of my own streams in which I can just do numerous takes until I get it right.
Twitch channel is here if anyone's interested:
https://www.twitch.tv/esamarathon