• Content count

  • Joined

  • Last visited

Community Reputation

37 Excellent

About weissel

  • Rank
    Rocketry Enthusiast
  1. Works on Windows at well (if you install cygwin, which I recommend for everyone who occasionally need the power of (or is comfortable with) a real, sane command line ) Or you can simply use your main RAM as the RAM disk (if you have enough RAM). Then the speed of reading and writing is basically the same speed as your RAM, vastly faster than an SSD. And then the drive is just a s power hungry as your other RAM[1]. And sequential reading from a HDD does need minimal seek time, so if you, say, save an image of the RAM disk (or otherwise structure the data and file layout to be mostly sequential), the loading of 8 or 16 GB should be on the order of 40 - 110 s (3/4th of a minute to just less than 2 minutes), assuming 150-200 MB/s from the HDD and uninterrupted bulk loading (no other disk activity going on). [1] that's not totally true ... Laptops in sleep mode often purge all the RAM they can, i.e. everything that can be recreated from e.g. the disk(s), which means amongst others any cached disk data by the OS (with dirty pages being written out first, of course) so they can shift what they really have to keep into the lowest number of DRAM chips,. Since DRAM needs power (the cells have to be refreshed IIRC 2-3 times per second), less chips to refresh means less power used, longer battery life, ...
  2. Yet another possibility is to use a better SSD-cache and/or RAM-cache system that learns what is accessed when and can thus pre-cache and also return data that has been read recently. This is most important for small files as the access time on spinning rust (HDDs) is vastly longer than the actual read time. And here the very fast access times of SSDs come in handy. Or even the (again) much much faster access times of RAM. On the other hand, HDDs are best when reading large chunks sequentially — movies, large resource files for games … and if they have to seek to a new fragment every 500 MB, that is not that bad (compared to every 4 KB) and therefore needs caching least. And it would take the space of hundreds of small files, too, so the cost of caching is highest and the gain is smalllest — which is a happy coincidence. Theoretically your OS should already do proper caching with all of the currently unused RAM, but with Windows … let us just say the number of 3rd party caching solutions out there is telling.
  3. I’ll try. It’s hard to take out the technical details because then you have to “simply believe me” instead of being able to reason — and tell me where I have gone wrong. In very short: SSDs are very different to HDDs internally but are talked to as if they were HDDs, and despite a lot of smarts used internally by the SSD to keep stuff running (and some help by the OS[1]), heavy use can slow down an SSD a lot, especially older ones and nearly full ones. The only “cure”, if that becomes too bad, is to basically fully reset and empty out the SSD via “secure erase”. After which you get to find out if your backup works. And with the same lifestyle, I mean usage patterns you get similar results, and the SSD is not getting younger. But no cure before diagnosis: do get yourself an (SSD) speed test program — there should be plenty around for free — and see how fast random 4kB and sequential reads are, and compare that to what the manufacturer stated. You might also want to track the queue length (how many commands on average are waiting for the drive to have a slice of time for them) and transfer rate (how fast reads/writes are and %idle time (if the drive is all running flat out or if it has some space in-between) while starting up KSP. This will help to find out if the problem is mostly the drive or not in first place Any clarity left over? Just ask and I’ll muddy it for you! [1] TRIM can tell e.g. an SSD that these and these sectors are only holding junk data now and do not be copied or preserved. Since copying stuff around is necessary, not needing to copy junk data is a boon.
  4. Quote from the FSF needed, naming KSP and mods explicitly. Otherwise it is just your interpretation. You can do whatever you like if you break no laws, licenses and contracts. This includes robbing a bank. Of course you might have to own the bank and all the money therein first … or it might have to be a movie set and actors. And of course you can comply with the GPL, for example the same way you can comply with an ARR. To claim otherwise is disingenuous. How do you license anything “incorrectly”? How does that even work? And where, except in your mind, is the user not allowed to distribute binary versions if they fulfill the terms of the GPL? In your mind, yes. So … how again do they distribute a binary form of a mod or derivative and respect the ARR license? Well, you are not any higher than the contradiction level yourself, yet you think you are on the peak. Which law school did you visit again? Moving goalposts a lot lately? The discussion is about the written text of a license and the interpretation thereof. But well, the original intent was to increase free (speech, not beer) software. Writing GPL mods actually furthers that goal. As to KSP vs Linux: you can write GPL code for Symbian (what Nokia used to use, back when). Or any other strange, different, one-of-a-kind niche system. How about the Dreamcast, for example? So your point was? And what if you ran, say, a GPLed Atari 2600 program — on an emulator (closed source), within a virtual machine on a virtual computer (say, like Amazon’s cloud computing servers)? Which parts would you have to package if you wanted to pass on the Atari 2600 program (which you got under the GPL and modified yourself) in binary form? The emulator? The virtual machine the emulator runs on? Inquiring minds want to know!
  5. How can you violate a license you are not bound to? If a judge uses the internet, do they violate the terms of parole they routinely give: "No using of the Internet!"? How about you simply say that in practical terms, an exception for KSP will not hurt (as the mod is supposed to run with KSP in the first place[1]) and will make it so much easier for others to carry on the torch should you falter, or build upon your work, or fix bugs, ... That is a sensible argument, easy to understand, non-confrontational and not aggressive, not attacking them and telling them they are stupid, too stupid to see. Speaking of honouring a license: with GPL you seem to argue to take upon oneself the same limitations the license passes on to others. Now, what happens if we apply that not to the GPL, but some other license, say ARR? Do you want to plead special circumstances for each single license? Or would it be better to find a legal theory that is self-consistent, say, like "Magic: The Gathering", and not randomised like lottery numbers, without rhythm or rhyme, like, say, the Windows-APIs? If the developer releases the code under the GPL but bundles a binary file, too, what stops you from using and distributing the code (with or without alteration) but *without* bundling a binary? It's not even necessary that your work compiles or runs --- or be even meant to do that --- it still can be under the GPL! That is exercising rights the GPL gives to you, even if it's not all you want --- in fact that an exception is required shows you want more rights than the GPL can give you! Assume you had a "integrate into your work, copy and sell in binary form only" license for some library you use. You may have paid lots and lots for the license. And now, because you added some GPL code, you cannot exercise the rights the GPL would give you if there wasn't that library, nor the dearly bought rights the library license gives you. Does that make licensing meaningless? [1] And even if T2 would pick up the mod and bundle it, they could, but would have to pass on the source and build tools as required by the GPL, except they do not need to GPL KSP.
  6. If you have the RAM (24 GB minimum, I'd say), pre-load KSP into a RAM disk. RAM is many times faster than SSDs. If you use SSDs, the only way they know the computer has "thrown away" data on the HDD they emulate is when the computer actually overwrites a block --- or when it is being told by the computer. The TRIM command tells the SSD (ranges of) blocks that carry no useful data anymore, so they need not be copied when the data has to be moved to a new erase block (128 kB usually) --- you cannot overwrite on flash memory, you need to wipe a whole erase block, which means find which erase block to use next (wear levelling is very important for a usable life time) copy all the information not known to be useless ('overwritten in the logical 'HDD' emulation or named via TRIM) to a reserve erase block. wipe the erase block and put it in the "ready to reuse" queue. This causes a more and more "fragmented" SSD, with many blocks partially full, despite the garbage collection the SSD runs in the background. Erasing is slow. Writing to a flash cell where the logic does not only expect 0% and 100%, but, say 0%, 25%, 50% 75% and 100% is *really* slow as you have to poke the cell that it only holds the charge you need it to hold. This means usually several pokes and reading the results after every poke. But as it allows you to cram more storage into the same flash cell(s), consumer SSDs tend to have that. (Physically the cells are identical, and it is not unheard of an SSD that uses on/off for critical internal data, like where in the physical flash the data for the virtual block (that makes the computer see a HDD) is, or even as cache during write operations, speeding up writes (until the cache is full) and committing them to the many-values-cells in the background. Many consumer SSDs are not graceful when TRIMs are sent for each block, as they are invalidated. However, most are amenable to a, say weekly run, program that marks whole ranges as "free for reuse" via TRIM. If your SSD does not do TRIM, consider a backup, a total wipe and reset to factory conditions for the SSD, and then restore what you need onto the SSD. This cleans up any internal fragmentation, when 128 kB erase pages are half filled and a lot of extra erasing and writing is needed to keep "data" the file system has abandoned. (-> Write Amplification). There are also 'prefetch' programs for Windows and Linux (and I'd be surprised if they didn't exist on OS X), which intelligently try to pre-load and cache what likely is asked for next --- they partially overlap with RAM caches which keep or toss data in RAM as the algorithm says it's worth it or not.
  7. Do I understand your logic correctly when I say "If the creator publishes their work under the GPL, they are bound by the GPL. That means amongst other things, if they publish compiled binaries/dlls bound to KSP means the creator --- as spelled out in the GPL license --- must also deliver the code required to build and run the work." Is that materially what your logic is? Yes? OK, then let's see what happens ... "If the creator publishes their work under GPL ARR, they are bound by the GPL ARR. That means amongst other things, they may not copy, pass on, modify or publish their own code --- as with ARR --- nobody has the right to do so." This is the exact same logic, and the result is obviously nonsense, therefore the logic is clearly wrong. So please stop applying faulty logic, it gives you worthless answers. Or to quote the GPL FAQ: Is the developer of a GPL-covered program bound by the GPL? Could the developer's actions ever be a violation of the GPL? Strictly speaking, the GPL is a license from the developer for others to use, distribute and change the program. The developer itself is not bound by it, so no matter what the developer does, this is not a “violation” of the GPL. (emphasis mine). Can we now all agree that this is straight from the horses' mouth and authoritative on this topic, and the developer (assuming he is not including/using/... other people's GPLed code) can distribute compiled binaries, with or without source code, with or without all necessary libraries and build scripts and there is no GPL violation problem for him? Also consider how dual licensing could work under your theory. The GPL does not prohibit any kind of use. Copyright law does. The GPL allows you to legally perform certain actions --- which otherwise are disallowed by copyright legislation --- if you fulfil certain conditions. That is all. You are not forced to accept the GPL license. In which case you do not get the additional permissions the GPL confers. You asked T2. You forced them to take notice. You force them to make a decision. No, that does not affect T2, not even PR wise, not deciding if and what to answer you ... suuuure. In other news, if you kill someone, that literally does not affect the police or the courts in any way whatsoever. That is the one true statement you made. The EFF has literally no way to shut down modders or T2. And I cannot see any reason why they would even want to. I shall just point out that you yourself seem to have a very sketchy understanding of software licenses, at least when it comes to the GPL and how it interacts with copyright. And just because others get it wrong doesn't mean you get it right. Anyone concerned over licenses should pay for and ask a competent copyright attorney. (Which I am very certain you are not.)
  8. Better Than Starting Manned Revival?

    CC licenses are not good for programs — no need to include the source code — but probably good for textures, meshes, models and art.
  9. Oy vey. The GPL gives you additional rights that you do not have due to copyright laws. The whole linking-or-not with KSP is spurious unless you want to distribute (convey, make available, …) a GPLed work as an integral part (linked, deeply interacting, not just using config files and command line arguments) of some not-GPL work. Since only T2 may allow any distribution of KSP, a discussion about distributing KSP and a GPLed mod together is moot — we are not allowed to distribute KSP in first place! Now, 2 mods that interact deeply — sorry, MM does not need to apply here, it just reads it’s instructions and does it’s thing — with one of them GPL (and not solely yours[1]) and a not-GPL compatible mod, say ARR, that you cannot distribute that way. If your, say, ARR mod depends on/needs the GPLed mod which has no ecxption for your ARR mod, you cannot distribute the ARR mod. [1] you can trivially add another license to a mod where you have all copyrights, just ask youself “May I distribute it as ARR, too” and answer yourself with a “Yes, sure, no problem” …
  10. Better Than Starting Manned Revival?

    ARR means the mid/game/program dies with you — or even without you if you have to abandon it. And that is a pretty sad thing if it hits good mods. (You can always add a “now new versions for x time yadda yadda and the licence reverts from ARR to <a licence that lets people keep it/a variant under a different name/... alive>.) Oh well.
  11. Better Than Starting Manned Revival?

    If I only had the time and the spoons to make a spiritual re-imaging of BTSM... Of course we cannot (without FC's explicit, written permission) copy any of their works, as it is ARR --- all rights reserved. And I understand why, though I wish there'd been a clause that if there's 1 or 2 years without a BTSM release or something, the licence would change to something that allows someone to keep the mod alive. Maybe under a completely different name and making clear that this is not BTSM ... but it is as it is, and BTSM was great and taught a lot. What we are allowed to do is to understand the reasoning for the decisions made. Like gating technologies to have something to work towards, say, radial decouplers, fuel lines, solar cells, better engines (the aptly named "Death Wish" was ... fun. Frustrating as hell, but still better than solid engines (especially w/o changeable fuel level and thrust limiters), and made you appreciate --- and work towards --- a better LF engine.), ... Or the bombing missions ... something totally different from the usual fare. Or the fact you can only try a mission once, you fail, then you fail. Or how you would have to stack up missions to make money. Or how there always was a 'routine' mission to bootstrap you back to some money. Or that there were no dead ends in the tech tree, but you might have to be very inventive. Or, for example, how the limited amount of battery power you could carry would teach you that the Hohmann transfer is a very fuel-efficient and very time-inefficient manoeuvre, and ... encouraged you to think in other terms. (Cue back-of-envelope calculations where the minimum mass of fuel for a faster travel and batteries for the travel time was.) Or that the life support use in the lander can was bad and in suits abysmal, which gave you tight, but doable limits ... Yes, restrictions enable art and creativity --- and we love that (or we'd be playing in the sandbox, using the best engines and cheating ourselves into and out of Mohole and Eeloo orbit ...). And BTSM gave you just enough that problems could be solved, in several ways ... but not easily. Adding to that --- I had great fun playing with RP-0 and the unreliable "Ajax " engines (Taerobee), trying to make sounding rockets that would handle the failing of one or 2 of them in a stage gracefully. Something like that --- and parts getting better as they are used more and stuff --- might have a place in ... uh ... "Kart Mode" (K-hard and K-art and cart and Go-kart ...) or whatever the spiritual re-imaging would be called.
  12. Unless you are from the Duchy of Grand Fenwick and have their unbeatable army back you, you do not stand a chance.
  13. There are harder-to-get jobs: POTUS, Pope, King or Queen in a strictly royal bloodline system ... Choose Another Dream Music and lyrics copyright © 1998 Mark Bernstein
  14. Children of a Dead Earth: realistic space warfare game

    And now read the Atomic Rockets website and find out that "stealth" again is relative. Yes, you likely can use hydrogen or helium to dump heat and expel the waste (or use it for thrust, too) while being hard to see. And now imagine you were the guy in charge of protecting your planet/asteroid/system from threats like these "stealth ships". Are you going to have lots of small, cheap drones around that look for heat signatures? Sure you would. How about drones scanning everything nearby with Radar or Lidar or similar? Sure you would, even if they are more costly and visible to all due to their higher power use and active radar/lidar use. And if they are attacked, that in itself is an important alert signal. How about extensive surface systems, which can be totally stealthed under ground, both for observation and for counterattacks? As much as the budget can be squeezed and then some! How about organising a 24/7 watch on the brightness of all stars bright enough to be of use? No problem! We routinely do monitor star brightness of stars over months and years, detecting tiny dips of 0.1%, which shows us there's a planet crossing the star far away. All that is needed are some large aperture lenses (apparently the old 200mm f1.8 Canon lenses do great work for a very low price --- compared to specially made optics, at least) and sensitive sensors (current DSLRs have fairly good sensors, enough for that work) and some software for tracking and finding the stars on the photos and measuring the brightness. Which we all have today. I am pretty sure you'll find some other ideas --- even a very sparse, but wide hydrogen stream at 9km/s (and 22K) should be detectable with matching active sensors. I am thinking of something not quite dissimilar to Michelson-Morley (trying to detect the Aether which obviously failed due to non-existent Aether --- but that hydrogen cloud exists ...). All that being given, do you see a role for spy craft of that kind? And if so, will they (at 190m/s per day) be manoeuvrable enough for any kind of battle where they are detected? (I guess no. Once they are detected in any way, any dumb kinetic weapon will take them out, they cannot dodge at all. So yes, you may have some stealth in space --- given very special circumstances --- and now tell me, what can you do with it?
  15. Better Than Starting Manned Revival?

    "Better than <questionable design decision>" has become an in-joke for FlowerChild's mods. Better than Wolves, Better than Giant Bees[1], Better than Sentry Turrets (which was the best he could come up with re Rimworld). For a "meaningful, slow progression" starting with valuable and expensive Kerbals is not the right thing. But then KSP's Career Mode is a tutorial for beginners, not a challenge even for veterans. And there I can really see the "better than" being true --- obviously if you have different goals, your mileage will vary a lot. [1] 7 Days to Die, a craft/build/survival game with zombies and zombie hordes coming to kill you. They had Giant Bees as a 'flying enemy' back then, so you'd need some sort of roof, too. The game was too easy for FC. They added a shortness on resources, resource quality by biome (the more dangerous, the more rewards --- instead of "the higher your level and relevant skills, the higher level the rewards), slower progression, reduced the fairly OP crossbow, etc etc etc and you'd avoid zombies --- or lead them away from where you wanted to search for stuff --- because zombies now are dangerous. A few versions later, while not perfect and still having some less than perfect things, settings to hard (zombies ruin always, shortage of everything, no loot respawn, no air drops, ...) FC found the gameplay hard enough for their liking and discontinued the mod. (Note that FC is/was a game designer)