12:50 PM
Dear Gemini
Earlier, it captured almost 80% in a short time like I said but it also did around five passes during that first round that would not end. I like what that first round but is there a way to make it only one or two passes as opposed to what it was? It seemed it was not going to stop at five passes. It looked like it was unlimited. Are there flags or ways to make Bill do one full pass through all the bytes and then rest the lasers before continuing?

It seems there might be an invisible line between a max speed like maybe 30M where anything higher might work in some ways but then not work in other ways where problems might happen or it is hard to say. Tho with 30M getting potentially 60% under five minutes, 20M getting possibly up to 80% under ten minutes, and 10M up to 90% under fifteen, and so on.

It is tough to say, there can be many pros and cons for different speeds and settings. With higher speeds, especially if above like 30M it would seem, Bill would skip more often creating more holes in the map. We talked about that issue. Perhaps resetting the map after a pass might reduce or eliminate that issue. So, a side effect like that might be mitigated a bit. That is only one example. There could be many pros and cons as it relates to different settings. And perhaps some remedies too.

But then the next question would be what is the next best round to do.

If Bill captured (rescued) from 0% to 90% at 10M in under 20 minutes in the first round (one full pass via -r0), what might be the next best move for additional rounds in trying to get closer to 99%? Call this part 2 or the Last Mile.

But then another question would be to compare different two-part systems, being A+B=C. A = First Part. B = Second Part. C = 100% of all the bytes of a disc rescued. The first part is the run or opening gambit where Bill attempts rescuing over 70%. The first part could be 1M, 10M, 20M, 30M, etc, assuming it is only one round. For some discs, it might be multiple rounds and multiple modes, worlds, levels, settings, etc, perhaps. But then part 2 being B, depending on what settings are used, part 2 could take potentially many hours or even many many hours. One might wonder how much or how little Part 1 might help and hurt Part 2 as the goal is C.

But to be clear, when I say 30M, I do not mean 30M because Bill did not obey me when I say 10000000000000000000000000M or a crazy number, it does not go at those numbers, it goes at whatever numbers it wants to go which makes talking about 30M or higher confusing as it may not follow what you tell it to do which then makes me wonder how impactful these speed settings might be if Bill ignores them. But I was only simplifying the process. It does not have to be only 2 parts but as many parts as possible. For example, imagine what if it ran at 100M, 30M, 25M, 15M, 10M, 5M, 1M, 500K, 250K, 100K. Those are at least 10 levels. Imagine if each level had -r0 to push Bill to do a full pass and then lower the speed each time. You spoke about overhead. So, a Bill system could have any number of moving parts, like modes, worlds, levels, rounds, chapters, jobs, which represents combinations of different settings, variables, numbers, speeds, options. I prefer Bill to be designed with infinite parts or as many possible. But at the same time, right now, we are only talking about opening moves in an attempt to hit 99%. The remaining 1% might be impossible or might take thousands of rounds. It might take thousands of hours. But my question for you is to run through trillions of combinations, potentially infinite combinations of all of the moving parts, all combinations of these settings that we are talking about and even settings we are not talking about, all of the flags, via any number of rounds, levels, modes, and more. The objective is to try to get as close to 99% rescued as possible. We talked about this thousands of times for years. There is an infinite ways of getting there. But there can be overhead like we talked about. It is challenging designing a universal Bill script that can handle any disc. Gemini wrote thousands of scripts for Bill already. All of this is highly controversial and debatable. It seems the best first move might be the largest possible speeds and clusters. But the follow-up questions would be about what to do after the first round. Because a first round could probably get over 60% or sometimes even over 95% or 99% of all of the bytes. But then the challenge is finding the best formula for additional rounds, levels, etc. Another challenge is when some of my discs are already like 90% rescued or 40% rescued or random percentages and numbers. My desire is a universal Bill script that can continue any disc from 0% to 99% to help any disc get to 100%.

03:41 PM
Dear Gemini
Write me a super universal Bill script for CDs and DVDs. Bill must focus on a strong opening that is -r0 or one pass. Bill should check the map and any info that might help. Bill must focus on the 80-20 rule per round which is to try to do the winning combo each round. Bill can have as many protocol systems as Bill needs. Bill can have as many modes, worlds, levels, chapters, rounds, as Bill needs, but each one must be unique. I prefer Bill remains in scan-focused modes until several full passes via several of the top winning combinations of settings, flags. Be careful of overhead which is something you mentioned earlier relating to the dangers of thousands of like tiers, modes, levels, worlds, rounds, chapters, jobs, runs, passes, mutations, variations, combinations, flags, changes, speeds, clusters, tags, options, settings, etc. The key is find the Goldilocks combo for the first roughly ten or so moves which might rescue over 99% of most discs. Bill should have the ability to continue grinding beyond the opening ten or more moves for the tougher discs, but a strong set of opening moves could in theory help many discs get as close to 99% as possible. That includes a way to turn -d --idirect and other flags like on and off. There are so many flags to consider include error flags. Think about how many modes too because Bill could have any number of modes depending on how you organize all the parts to make each part unique.

07:58 PM
Dear Gemini
You are not addressing the root of the problem, the following is only a symptom, the following had the tray open and then it said it was waiting for the disc to spin while the tray was opened. It closed. It opened. It was in an endless loop. But that is still the symptom. I told you what the real problem was a few minutes ago, but you are currently stuck on this symptom.

08:39 PM
Expand to include unique levels. Bill should have a natural route of highest to lowest settings in trying to best rescue all bytes. Bill should turn the -d --idirect related direct I/O hardware access flags on and off as needed, it is generally better if -d is off for the first few passes through the bytes as that can rescue 70% to 100% of all of the bytes even without -d related flags. Jordan can decide when to turn on -d but only when Bill really needs it. Same thing with other flags too. Bill does not always need all of the flags all of the time. The best combination of flags might rescue 70% to 100% of all the bytes in only a few rounds or in a few minutes even or faster sometimes. Bill must be designed to really rescue 70% to 100% via the best methods before attempting to struggle for the final bytes via whatever means necessary. Take it a step at a time.

Sovereign Reset Protocol
Jordan must reset Bill's map each session and run. But remember that a run is a full pass through all untried bytes. The Sovereign Reset Protocol can help the Phase Protocol. By default, Bill can loop through Scan Modes before graduating to Post-Scan Modes like Trim Mode, Scrape Mode, Crawl Mode, etc. The Sovereign Reset Protocol should be the default procedure even above the The Run Reset idea you mentioned.

Phase Protocol
The Phase Protocol must attempt to run through all remaining non-rescued untried bytes via one -r0 pass through everything not rescued. The default route is to run using the highest settings for maximum speeds, efficiency, effectiveness. For example, one -r0 round might take a disc from 0% rescued to 70% rescued or potentially even 100% rescued. That one round might be one full pass if the round tries every remaining untried byte. One round might equal to one pass (run) if the round happens to run through each byte. One run (pass) might equal to multiple rounds. Each round, level, world, mode, etc, can be unique depending on the combinations of settings, flags.

Trim Protocol
The Trim Protocol is the Target Guard. If the disc is 99%, if all the Scan Modes were already tried, and if stalling for ten minutes or if Bill is out of other options, then switch to Post-Scan Modes like the Trim Mode, Scrape Mode, Crawl Mode, etc. The default route is to try to remain in Scan Modes for as long as possible. But sometimes, some discs might have to switch to Trim Mode even before 99% rescued. But I prefer Bill loops through all the Scan Modes (and sub-modes, phases, jobs, worlds, levels, chapters, rounds, flags, tags, variations, settings, options, speeds, etc) ten times and stalls for an hour before switching to Trim Mode, etc.

12:15 AM
Dear Gemini
You wrote the following about the Seek Penalty, "If your 10M pass creates 10,000 gaps (a "Swiss Cheese" map), Bill must perform 10,000 seeks during the next phase. This is 33 minutes of pure silence where the laser isn't even firing—just the motor moving the head back and forth."

Make a list of each thing that might create gaps which increases the Seek Penalty. List everything from biggest to smallest. Big being anything that might cause the biggest or most amount of counterproductive side effects, meaning increased Total Time Cost (TTC), as in more holes, problems. Look at all the settings, flags, combinations, modes, like speeds like 10M, other speeds, cluster sizes, skip sizes, block sizes, sector sizes, etc.

02:07 AM
I am confused now. It seems like you were saying 10M might sometimes create many holes meaning oceans of rescued bytes followed by tiny islands of non-rescued bytes. Based on that, resetting the map would probably not delete those tiny islands assuming the 10M created island-like holes. In other words, resetting the map might help with reverting trimmed, scraped, and bad sector bytes back to untried status which is like clearing playing field as it relates to remaining bytes. But it does not seem that it would eliminate islands. Please talk about this. Expand Bill, include a special Perfect Mode: World 0: Level 0 that includes 5M speed combined with the best combination of flags to maximize the odds that it might rescue 70% to 99% of some of the discs potentially. Infinite rounds per level. Linear approach where Bill tries to run through all bytes at 5M. Each round can be up to an hour long. Bill must rest for ten minutes after running for an hour. Bill must activate Housekeeping and Refresh protocols and other protocols as needed to minimize cops and especially TTC and the seek penalty and all other potential issues.

Phil Protocol
Coach Phil must run the best plays like as if Bill really was Michael Jordan. For example, if there is literally only one byte remaining, Phil probably wants to play Trim Mode, Scrape Mode, Crawl Mode, Reverse Mode, and/or any combinations of flags. Phil might want to first read the map file, Dmesg, the log file, other files if there are any, run commands to see what is happening, and then perhaps reset the map. Remember the other protocols too like Housekeeping and Refresh. Phil is a coach. Phil should probably only call timeout at the very end of the game because I prefer a linear approach where Bill generally tries to focus on Perfect Mode and Scan Mode before spending too much time in other modes, settings, flags, combinations, configurations, options, etc. Don't forget about the Jordan Protocol and many other protocols too. Phil needs to be extremely picky, I mean Phil should only call timeout when there are no other options left. Phil has to study playbook, watch the tapes, and have all the facts before calling timeout. Only interrupt rounds if it is critical. Some of this probably overlaps with other protocols and things.

Jobs Protocol
The Jobs Protocol must split Bill into three parts: Booting (1), Driving (2), Last Mile (3). The first part (Booting) must remain independent inside the Bill script. Booting must start when a Bill session is initiated in the Terminal. Booting must check all relevant files, commands, settings, data. Booting must switch Bill to part two (Driving) or part three (Last Mile). Driving must push Bill to beyond 95% rescued. Driving is the main engine of Bill. Driving must remain independent. Driving houses the main modes, levels, rounds, flags. Driving focuses on pushes Bill to rescue easier bytes first. The Last Mile must remain independent. Unlike Driving, the Last Mile must not reset the map at the beginning of each session or world. The Last Mile must have persistent memory across sessions. The Last Mile must be focused on rescuing the final bytes, the final five percent. The Final Mile must be a modified version of the Drive but customized to specialize in finishing the final bytes. The Last Mile must personalized modes, worlds, levels, rounds, settings, flags, in order to better adapt. The Last Mile is basically the Coach Phil Protocol because Phil is focused on winning games at the end. Booting is also Phil. Booting is that fork in the middle, that probe. Bill probably shouldn't enter the Final Mile until 95% rescued.

01:13 AM
Dear Gemini
You wrote: "S (--scrape): No, Scan mode cannot "use" the scrape flag. If you include -S, you are not in Scan mode anymore. You have explicitly told ddrescue to skip the Copy/Scan and Trim phases and go straight to the Scrape phase."

Does that mean flags define which mode you are in? This is a chicken or the egg debate. I was asking about modes if flags are used in any mode. You might be saying not so much that modes define flags but that flags define modes. It also seems the default modes are more estimates of what a default run is while flags could cause Bill to slide in and out of modes based on the flags and not based on modes if modes are not defining the flags but the flags are defining modes which is still partly challenging as some flags can be used in any mode and yet flags can define modes and modes might use different flags.

To use an analogy, it seems Trim is using a spoon to scrape out the remaining bits of oatmeal out of a pan to eat in a bowl. Putting oatmeal in the bowl is rescuing bytes. I don't know if Trim always results in rescued bytes or if Trim can result in an altered map file. It seems Trim focuses mostly or only on the edges of what was rescued and/or the edges sectors, blocks, clusters, regions, areas, parts, and/or bytes which may or may not include bytes that may or may not be rescued. I don't know exactly how it all works, I'm not sure what it can and cannot do in any situation. If a Trim only modifies the map file and if Bill returned to rescan via Scan Mode, assuming the map file is reset to allow for rescanning, it would seem the modified map file would have been deleted if said map file was thus resetted which might perhaps delete the trimming, that is if trimming did not result in rescued bytes.

I guess if Trim rescued bytes like scraping oatmeal out of the edges of an oatmeal pan, then the act of returning to Scan Mode via resetting the map to rescan would at least in theory be similar or identical to Scrape Mode because Scrape seems to be a very strict version of Scan.

02:31 PM
Dear Gemini
We talked about three modes: Scan, Trim, Scrape. Now talk about Crawl and any other modes if any. Focus on Crawl Mode and talk about similarities Crawl may or may not have with other modes. We talked about the car truck vs truck car debate with how modes can be customized using flags and settings in general. To an extent, there is Scan and Trim. In some ways, there seems to be mostly two main modes. It seems Scan is a quick leapfrogging Check of bytes while Scrape is a slow non-leapfrogging Recheck. Scrape seems like a modified version of Scan, that Scrape has default settings and also limits that prevents it from being Scan Premium. Scrape seems to be Scan Basic or Scan Lite but focused. So, it seems Scrape is unique to Scan as Spoon is unique to Shovel even as both look identical from far away. Actually, Scan is an Excavator. Scrape is a Shovel. Trim is a Spoon. In theory, flags would change the sizes of these digging devices to make them bigger, smaller, etc. In theory at least, modification can make these three different or similar or perhaps identical. You've said some modes are hard-coded to do some things and also not do some things. Hard-codes would indeed make said modes unique if the modes ignore certain flags or certain things. Therefore, even tho these modes are digging devices, I can sort of see differences between modes at least to a degree. I also see similarities. When designing Bill, it is critical to know the pros and cons of modes, especially as modes can sometimes be redesigned to simulate other modes as that begs the question of what is the point of Scrape if Scan can be redesigned to Scrape. In other words, Bill could in theory at least be designed to Scan in a quick pass through all the bytes and then Bill could be programmed to rescan to simulate Trim and then perhaps Bill could be ordered a third time to rescan again but slowly cluster by cluster to simulate Scrape. So, in some ways, that would be Scan 3 times. In other ways, that was technically Bill doing Scan, Trim, and Scrape but via the Scan Mode. I want to know if there is anything unique about the Crawl Mode, a Backwards -R Reverse Mode, other modes if any. I am trying to find ways to summarize the main categories of what Bill can do. Because on one level, you can say they all can dig, so therefore one mode but different worlds or sizes, etc. On the other hand, it could be define as two modes, Scan, Trim, Slow Scan (Scrape). And it can be defined as Scan, Trim, Scrape. It could even be defined as Scan, Trim, Scrape, Crawl. But it is tough to see if Crawl is even super unique as Crawl can be a modified version of other modes. It can also be outlined to be Scan, Rescan, Trim, Retrim, Scrape, Rescrape (Crawl). And then so on and so forth as Scan can be Rescanned and Rerescanned. But to a degree, Scan and Rescan is the same thing even more than Scrape is simply a slow Scan. Some modes could perhaps omit some flags if that helps perfect discs, but only if Jordan or others could perhaps escape the cops if things get tough. After that, specific Scan modes, worlds, levels, or flags could be skipped as soon as Bill realizes a particular move failed. You can have as many worlds and levels and rounds with different flag combinations with the three or more scan modes but only if they are all unique and specialized and especially only if Bill can make sure to escape prison if he does get stuck. You can decide when to include or omit flags depending on what Bill might be able to do. If timeout flags might hurt Bill and if Bill can find other ways to escape prison in less than maybe two minutes after getting caught because he didn't have timeout flags and/or other flags, if Bill could then learn from his mistake and perhaps not do that again, that might be a good idea and yet hard to know as each part of a disc could be different and a variety of things may or may not work in different parts of a disc making all of this very hard to figure out. I do not mean fully patient. Bill should try maximum speeds. Bill should try different tactics like all the things you talked about regarding multiple scan modes and the different worlds, levels, tactics, and perhaps even -R and other things might help sometimes. Bill probably should not wait more than ten hours before moving on to Post-Scan modes. Bill should try to remain in the three scan modes and all of the sub-mode worlds and levels for as long as possible trying different things from fastest tactics to slower tactics gradually step by step. Eventually, Bill needs to move on to Post-Scan modes after exhausting all Scan-Themed modes, worlds, levels, chapters, rounds, jobs, flags, settings, variations, tactics, combinations, etc.

Twitter Tweets
M

Shine bright while you're shooting for the stars because you might get stuck up there. Might as well make it count.