[X] My Assistant

The -12V rail is good for 300mA to 1A or so depending on the supply. I think I've seen 1.5A on an old server PSU before.

the 1.5A one probably was using an LM7905 or LM7912 regulator.

My old Dell OEM PSU uses one of those.

Edit:

Now using a two button Logitech (AT&T branded) PS/2 mouse just because i like the way it feels in my hand. Got middle button emulation enabled and turned on the 'scroll method' where you hold the 'middle button' and drag the mouse to scroll. Probably won't keep doing this for long, but it's working surprisingly nicely.

"Fixed" my trackball's bearing which was slipping by propping the axle up with a paper clip. Works for now.

Trackballs are fun. It's a shame they have so many longevity problems.

This post has been edited by dragontamer8740: Apr 19 2022, 04:42

I forget if I posted this already. It's a shell script that creates bbcode for 'preview' sized imgur images that link to the full size images.

CODE

#! /bin/sh
# imgur link to forum post embed
while [ "$#" -gt 0 ]; do
  FILENAME="$(echo "$1"| sed 's!.*/!!g')"
  PREFIX="$(echo "$1"|sed 's/'"$FILENAME"'//')"
  FILEEXT='.'"$(echo "$FILENAME"|sed 's/.*\.//')"
  FILEBASE="$(echo "$FILENAME"|sed 's/\'"$FILEEXT"'//g')"
  RESULT="$RESULT"' ''[url='"$1"'][img]'"$PREFIX""$FILEBASE"'g'"$FILEEXT"'[/img][/url]'
  shift 1
  if [ "$#" -eq 0 ]; then # print results when finished
    echo "$RESULT"|sed 's/^ //'|tr ' ' '\n'
    printf "$RESULT"|sed 's/^ //'|tr ' ' '\n'|xclip -selection c
  fi
done

Example usage (note: example image is nsfw):

CODE

thumbify https://i.imgur.com/jKCdI7I.jpg

output:

CODE

[url=https://i.imgur.com/jKCdI7I.jpg][img]https://i.imgur.com/jKCdI7Ig.jpg[/img][/url]

if you have xclip, the resuts will be automatically put on the clipboard. Otherwise, you can copy-paste the output. It also will work if you give it multiple image links at once.

This post has been edited by dragontamer8740: Apr 19 2022, 17:14

I actually have a question for once. Maybe someone here can answer it?

Background:
I have a laptop maxed out at 8 GiB of RAM. It's running Debian Linux, and I'd like to try something somewhat clever to make photoshop work better since it uses two SMR hard disks (which is bad for scratch disk usage).

The Question:
Since scratch disk partitions apparently have to be at least 6GB, Is there a way I can make, say, a 1GB ramdisk (tmpfs) and a raw disk image somewhere on my HDD, and get Linux to first populate the ramdisk and then overflow onto the disk image only when needed (to avoid unnecessary disk I/O)? I'd like to be able to make files that are contiguous over the two volumes. It sounds almost like what LVM would do, but IDK how to tell it to prefer one volume until it's full.

Alternative Ideas:
My other thought is farming out scratch to my desktop over LAN with an ethernet cable and either NFS or SSHFS.
I can make any directory appear as a 'drive' thanks to wine.

I have done the reverse and forwarded my tablet digitizer's I/O to the desktop (via socat, since it is identified as an internal serial device) and used vnc to get a picture on the tablet display itself, but I'd like to avoid that.

As for why i don't want to just use the desktop, my laptop is comfier sometimes, and it has a tablet digitizer on the screen.

inb4 "get another laptop that can take more RAM/is more modern" or "replace your HDD's":
No. Maybe once my summer job/paid internship is underway, but not now.

Another thought: if/when I replace my HDD's, it'd be really cool if I could make a hybrid MBR on the internal disk drive so that it can boot on my laptop but still support HDD's larger than 2TB. I'd just have to make sure all the boot files are somewhere within the first 2TB.

if I could find an SSD that was on a mini PCIe and presented itself as a standard SATA controller instead of M.2, that'd be awesome. My laptop has a spare mini PCIe slot but it doesn't support M.2 disks as far as I can tell. An expresscard one would also be useful.

This post has been edited by dragontamer8740: Apr 19 2022, 21:46

Usually the -12V is just a reference voltage rather than anything to be used for powering stuff.

Things like RS-232 use it, for instance. Normally the kinds of cables you see carrying RS-232 signals don't include a power pin.

How an old serial mouse gets power [www.epanorama.net] https://www.epanorama.net/circuits/rspower.html - that page also sort of explains the -12v rail.

checks out, probably explains why it didnt like powering my shoddy soldering station

The -12V rail is good for 300mA to 1A or so depending on the supply. I think I've seen 1.5A on an old server PSU before.

I'll have to work out where it gives up, I know I have some power resistors around somewhere.

UPDATE: found one, psu bailed instantly, dont think the -12 to +12 is something thats safety guarded against OC, even though it did nicely cut off when I was trying to power something else from it before. will have to see whats dead where another day. very much an "I'm a genius!" "oh no!" moment

big edited in response moment:

Background:
The Question:
Since scratch disk partitions apparently have to be at least 6GB, Is there a way I can make, say, a 1GB ramdisk (tmpfs) and a raw disk image somewhere on my HDD, and get Linux to first populate the ramdisk and then overflow onto the disk image only when needed (to avoid unnecessary disk I/O)? I'd like to be able to make files that are contiguous over the two volumes. It sounds almost like what LVM would do, but IDK how to tell it to prefer one volume until it's full.

off the top of my head all I can think of in terms of combining disks would end up in them preferring to stripe and avoid filling up one storage before the other. but one thing that comes to mind is a custom fuse driver as you could tape together storage however you want. but I doubt performance would be something to write home about and I also kinda hate fuse with a passion.

Alternative Ideas:
My other thought is farming out scratch to my desktop over LAN with an ethernet cable and either NFS or SSHFS.
I can make any directory appear as a 'drive' thanks to wine.

would work okay enough in all likelyness, I cheat a lot on laptops as well with just nfs shares, I've even (with varrying succes) ran games with data files on nfs shares

This post has been edited by cate_chan: Apr 19 2022, 21:55

one thing that comes to mind is a custom fuse driver as you could tape together storage however you want. but I doubt performance would be something to write home about and I also kinda hate fuse with a passion.

I have never liked Fuse. Only ever seen one or two actually good fuse filesystem drivers (sshfs and ntfs-3g).

[networked] would work okay enough in all likelyness, I cheat a lot on laptops as well with just nfs shares, I've even (with varrying succes) ran games with data files on nfs shares

I just wish wireless lan was even remotely as good. :(

I just made another one of these (not my pic) so I could get a reliable connection:
(IMG:[i.stack.imgur.com] https://i.stack.imgur.com/HlF9u.jpg)

I really want a crimper but also don't want to wait for it to arrive. Thus, zip ties.

This post has been edited by dragontamer8740: Apr 19 2022, 23:39

I just wish wireless lan was even remotely as good. (IMG:[invalid] style_emoticons/default/sad.gif)

I find 802.11ax can be good. Its quite annoying that they stratify the features and capabilities of routers and clients/adapters so much. AX seems to deal with environments that are quite noisy with multiple networks quite well, when other gens would shit themselves.

I find 802.11ax can be good. Its quite annoying that they stratify the features and capabilities of routers and clients/adapters so much. AX seems to deal with environments that are quite noisy with multiple networks quite well, when other gens would shit themselves.

I'd use AX, but even if I wanted to spend the money there are no totally free software™ wireless chipsets on the market. Just like for 802.11ac, last I knew.

I actually do have an 802.11ax wlan card that I could put in my desktop, but that's using ethernet still so it's doing well without it. My router is 802.11ac but I'm only running it in dual-band 802.11n since almost none of my hardware goes higher than that anyway. For 802.11n the Atheros 'ath9k' chips work nicely (ar9280 and ar9380 are my goto's). AR5BHB112 and AR5BHB92 are the cards i look for.

dual band 802.11n is enough to play HD h.264/HEVC mkv files hosted on my desktop on other devices without hiccups. It's just for things where latency and retransmission time really matter that it has shortcomings, like if you try to forward an X client connection over WLAN.

Does ax do better in crowded areas simply because there aren't as many stations broadcasting? I barely know anything about ax.

I hope I can get a nice 802.11ac card that doesn't need firmware blobs/works nicely in openbsd some day. We're just now getting 802.11n dual band.

This post has been edited by dragontamer8740: Apr 20 2022, 07:26

Does ax do better in crowded areas simply because there aren't as many stations broadcasting? I barely know anything about ax.

Depends which model router you get, not all of them have all the features and some are probably similar to 802.11n with a bit more bandwidth at the lower end.

The higher end models can use beam forming and vectoring to better target devices and minimize signal interference. AX can use a broader spectrum for networking and is better at allocating resource units from the available spectrum. It also can reduce latency dramatically by being able to differentiate its own network transmissions, which was what caused a lot of drop outs or latency on congested frequencies in the past.

With my setup, I get +5 to +12ms latency compared to 50-300ms over older standards.

The main problem is that not all routers have all the feature sets. The range is way better but that could be because i bought a higher end model with multiple antenna.

Since scratch disk partitions apparently have to be at least 6GB, Is there a way I can make, say, a 1GB ramdisk (tmpfs) and a raw disk image somewhere on my HDD, and get Linux to first populate the ramdisk and then overflow onto the disk image only when needed (to avoid unnecessary disk I/O)?

Maybe [github.com] mergerfs?

Maybe [github.com] mergerfs?

It also does NOT split data across drives. It is not RAID0 / striping. It is simply a union of other filesystems.

Close, very close— but no cigar, it looks like.

I still think I might be using this for a couple other situations to simplify my byzantine directory structure on the 8 year old Debian install on my desktop. Thanks.

Depends which model router you get, not all of them have all the features and some are probably similar to 802.11n with a bit more bandwidth at the lower end.

The higher end models can use beam forming and vectoring to better target devices and minimize signal interference. AX can use a broader spectrum for networking and is better at allocating resource units from the available spectrum. It also can reduce latency dramatically by being able to differentiate its own network transmissions, which was what caused a lot of drop outs or latency on congested frequencies in the past.

With my setup, I get +5 to +12ms latency compared to 50-300ms over older standards.

The main problem is that not all routers have all the feature sets. The range is way better but that could be because i bought a higher end model with multiple antenna.

Mine's a quad-antenna 802.11ac model. I guess the 1900ac is just a bit outdated, but honestly that's like almost everything else I own.

Hmm, looking at linksys' site now, it looks like maybe they discontinued the 1900ACS (and the rest of the WRT* series)... None of them are showing up on their online store for me.

If I get another router it'll likely just be an EA8500 or something for now. Doesn't look like any current 802.11ax router has good support for alternative firmware yet, and stock firmware on modern routers tends to give me AIDS with their "IoT" features.

This post has been edited by dragontamer8740: Apr 20 2022, 16:34

Dude fixed the N64 code. It was supposed to be one of the best running N64 games but it was ful of unoptimized code and bug work arounds.

This dude doubled the performance of the game on real hardware.

That's pretty cool.

And also a pretty massive waste of time.

I'm glad someone did it, though.

Someone did similar to get stable 60fps in Pokémon Red/Blue/Green/Yellow/Gold/Silver/Crystal. But in assembly, of course.

Also, just finished upgrading OpenBSD to version 7.1 on my backup X201 thinkpad (non-tablet).

And I was able to pass off a shell script I wrote in two days for personal use as a term project so I'm feeling pretty pleased with myself.

This post has been edited by dragontamer8740: Apr 21 2022, 20:12

That's pretty cool.

And also a pretty massive waste of time.

I think its pretty cool. I quite enjoy the games where people are able to go back and fix / finish them.

That is nice, but it's just a shame there's not a lot you can do about games that were rushed schedule-wise and just fundamentally are lacking in content (Phantasy Star II).

Also the English translation of Phantasy Star II is buggy as hell. There's even a spot where the music engine just hangs until you reset the game or pause and unpause. And that's an unavoidable point in the game, so I'm surprised it made it out the door.

For really popular games, though, yeah, it's good. Like bugfixes for Doom.

Edit/update: Had to take apart my oldest model M again; Noticed my numpad '7' key was registering just before the spring under it buckled. every time I take it apart I end up making things uglier inside. Ended up having a bizarre problem caused by the membrane again when re-inserting it. Used a small length of aluminum foil to fix.

Every day I am more tempted to try something drastic like turning my model M into a model F (capacitive sensing) with a custom PCB and controller. But I know full well I'd leave it half-finished for a year or some shit.

This post has been edited by dragontamer8740: Apr 22 2022, 07:16

don't think I got that far in phantasy star II.

Somehow, the mass-renaming tool I use managed to get some pages out of order in a manga I'm reading.
I don't know if there was a bug in the feature "reset the numbering when changing folder", or if something went wrong because I removed some files before applying the reset process, but when I re-extracted the files, everything was in order, and could successfully be renamed properly.
I hope there won't be more errors like that, still I wish I knew exactly what caused it, otherwise there's no reason to think I could prevent it from happening again, even if I switched tool/method (IMG:[invalid] style_emoticons/default/mellow.gif)

I use 'mmv' for all my mass-renaming needs, if I'm on a system where it exists. Otherwise I can write shell scripts to do it, but it's more annoying that way.

To zero-pad one-digit numbers to two digits and simultaneously change the numbers from the end to the beginning of the file names:
before:

CODE

$ ls
catgirl_0.jpg
catgirl_1.jpg
catgirl_2.jpg
catgirl_3.jpg
catgirl_4.jpg
catgirl_5.jpg
catgirl_6.jpg
catgirl_7.jpg
catgirl_8.jpg
catgirl_9.jpg
catgirl_10.jpg

to preview the change without actually renaming, use -n:

CODE

$ mmv -n 'catgirl_[0-9].*' '0#1_catgirl.#2'

Prints the actions it will take:

CODE

catgirl_0.jpg -> 00_catgirl.jpg
catgirl_1.jpg -> 01_catgirl.jpg
catgirl_2.jpg -> 02_catgirl.jpg
catgirl_3.jpg -> 03_catgirl.jpg
catgirl_4.jpg -> 04_catgirl.jpg
catgirl_5.jpg -> 05_catgirl.jpg
catgirl_6.jpg -> 06_catgirl.jpg
catgirl_7.jpg -> 07_catgirl.jpg
catgirl_8.jpg -> 08_catgirl.jpg
catgirl_9.jpg -> 09_catgirl.jpg

#1 is the first pattern to match ([0-9]), and #2 is the second pattern (the file extension, or actually everything after the first '.').
So, after running it without '-n', the files get renamed:

CODE

00_catgirl.jpg
01_catgirl.jpg
02_catgirl.jpg
03_catgirl.jpg
04_catgirl.jpg
05_catgirl.jpg
06_catgirl.jpg
07_catgirl.jpg
08_catgirl.jpg
09_catgirl.jpg
catgirl_10.jpg

But notice that since then you have to manually do catgirl_10.jpg, you could instead have done it in two steps to fully automate it ; the following should work for all one or more digit numbers, although technically this one will match things besides numbers as well if your files aren't as consistently named as mine for this example:

CODE

# first move the stuff after the first underscore and before the first 'dot' to the beginning of the filename:
mmv 'catgirl_*.*' '#1_catgirl.#2'
# then add a zero to the beginning of all files that only have one numerical digit before their first underscore:
mmv '[0-9]_*' '0#1_#2'

Thought: I like the slight rumbling/rattling ball mouses make when you move them. Oddly satisfying.
Also, I wish I could find more mouse models that have three wide buttons on top and a thumb-operated scroll wheel. I hate middle clicking with most wheel mouse designs. ESPECIALLY the ones that have the 'tilt wheel' thing.

This post has been edited by dragontamer8740: Apr 23 2022, 01:22

I made a small c++ program to swap 2 filenames.
It had been a very long time since I made an actual program in c++, had to rediscover some basics, and also in this case, found out by default a project would be made with C++ 14, when a somewhat recent version is C++ 20 (and I needed a feature from C++ 17).

Still, it made me realize this is actually more complicated than it seems: someone out to bully a developer could just have files occupying all temp names you can come up with just to cause the program to crash/fail/do something wrong.
I just took the lazy approach by using a single filename for the temp name, and if the name is already in use, display an error message in the console (not even tested (IMG:https://forums.e-hentai.org/uploads/post-2051615-1433431238.gif)), then the program is made available through SendTo of windows for easy use.

You know, in unix, there's programs (albeit technically not standardized) like `mktemp` that handle random file name generation.
Have another shell script related to file renaming. This one avoids collisions using the while loop without depending on non-standard programs.

CODE

#! /bin/sh
# mvswap: swap names of two files in one step
# mvswap <file-1> <file-2>
if [ "$#" -eq 2 ] && [ -e "$1" ] && [ -e "$2" ]; then
  SUFFIX="_temp"
  # make sure our temporary placeholder filename doesn't overwrite
  # any existing files
  while [ -e "$1""$SUFFIX" ]; do
    SUFFIX="$SUFFIX""_1"
  done
  mv "$1" "$1""$SUFFIX"
  mv "$2" "$1"
  mv "$1""$SUFFIX" "$2"
else
  1>&2 echo "Error: Provide two files as arguments, each of which must exist."
fi

I also have scripts to offset numbered filenames by a user specified increment. Those are trickier but still quite reliable.
I question why a mass-renamer program would require C++17 features when such software existed before C++ even did.

Also, I just landed a > $20/hr summer internship, so I'm finally going to have money for parts to fix some of my broken shit/to actually donate to E-H and a couple other places/people I've wanted to help for some time. Pretty pleased with myself.
Hope I don't hate the work.

don't think I got that far in phantasy star II.

As much as I love that game, I can't blame you.

It occurs because of a buffer overflow where the game tries to write too much text when Lutz is explaining why everything's going wrong to Rolf/Eusis. It's very close to the end of the game, right after you get all the items from the dezo dungeons and before you get teleported to the final dungeon.
If you are interested, I have a save file you can use to check it out.

Edit: I just managed to do an accidental ctrl+alt+backspace (which kills X). Never thought I'd see that happen.

This post has been edited by dragontamer8740: Apr 25 2022, 00:10