Notices by clacke (clacke@social.heldscal.la), page 60

Snake Girl Ellie (noelle@elekk.xyz)'s status on Thursday, 19-Apr-2018 13:24:59 EDT Snake Girl Ellie

Mastodon tip: You can reply to your own toots. This won't put your username in the toot, but it /will/ mark your reply as part of a conversation.
Among other things, this makes it easier for others to follow what you're saying, if you're posting multiple toots in succession; all they have to do is click one toot and it'll expand the whole conversation thread.

In conversation Thursday, 19-Apr-2018 13:24:59 EDT from elekk.xyz permalink Repeated by clacke
Champion AR-15 (skoll3@shigusegubu.club)'s status on Tuesday, 08-May-2018 22:28:56 EDT Champion AR-15
@pea @shmibs @Siedge @fdavidcl

In conversation Tuesday, 08-May-2018 22:28:56 EDT from shigusegubu.club permalink Repeated by clacke
NASA (nasa@apoil.org)'s status on Tuesday, 08-May-2018 22:03:00 EDT NASA

RT @Dr_ThomasZ: NASA is an agency whose very purpose implies that we *have* to take risks - do things that nobody has ever done before. Rea… source: https://twitter.com/nasa/status/994034972460814336

In conversation Tuesday, 08-May-2018 22:03:00 EDT from apoil.org permalink Repeated by clacke
Jezza™ (jeremiah@noagendasocial.com)'s status on Tuesday, 08-May-2018 21:59:26 EDT Jezza™
- Anastasios Iraniن
#Orthodox humor
@gooz

In conversation Tuesday, 08-May-2018 21:59:26 EDT from noagendasocial.com permalink Repeated by clacke
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 22:21:44 EDT clacke
- jaff
- dewb
@dewb @jeff Real enough.

In conversation Tuesday, 08-May-2018 22:21:44 EDT from social.heldscal.la permalink
Jake (jakebeamish@social.coop)'s status on Tuesday, 08-May-2018 18:45:21 EDT Jake

Are there enough musicians in the fediverse to do a #Mastovision, with instances participating and voting on a song to represent each instance? I like #Eurovision but the patriotism is a bit meh

In conversation Tuesday, 08-May-2018 18:45:21 EDT from social.coop permalink Repeated by clacke
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 22:12:39 EDT clacke
- Shellkr
@shellkr Changing the sizes of various tmpfses doesn't directly affect the amount of memory available to processes, it just changes how much memory is allowed to be diverted to some specific uses (temporary files, named shared memory -- which is technically the same thing but has a specific API because conceptually it isn't). Just like most use of memory, these can be paged out if they and other allocated memory don't fit in the available RAM.

The system had 6GB of memory, 4GB of RAM and 2 GB of disk, available for its various processes, cache, buffers and other kernel memory. This wasn't enough for what I was trying to do. My choices in that situation were:

1) Reduce the memory used by processes and process-adjacent kernel structures, e.g. log out of my X session.
2) Increase the swap space, the only way to add memory to the system while running.
2b) I could have achieved this with zram as well, but simply increasing the memory on disk was the obvious way. The absolute best way is probably to add disk swap, but also make sure zswap is activated so the disk is spared if possible.
3) Go out and buy more RAM.

Swappiness doesn't enter into it. If memory runs out, it has run out. At that point the system has evicted all the clean cached pages it could, and paged out all the dirty pages it could, before saying there just isn't enough memory. Swappiness affects what the system does when it's in a calm state and is not forced to swap out anything. With high swappiness it might choose to swap things out to release RAM for buffers and caches, or available RAM for future processes.

In conversation Tuesday, 08-May-2018 22:12:39 EDT from social.heldscal.la permalink
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 21:39:19 EDT clacke
- Sean R. Lynch ☑️
@seanl Ok, so there was some deeper story behind that barb. I understand your position better now. Thanks!

In conversation Tuesday, 08-May-2018 21:39:19 EDT from social.heldscal.la permalink
Quad 🌸 (quad@weeaboo.space)'s status on Tuesday, 08-May-2018 07:56:39 EDT Quad 🌸

me✏️irl

In conversation Tuesday, 08-May-2018 07:56:39 EDT from weeaboo.space permalink Repeated by clacke
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 11:27:28 EDT clacke
- Sean R. Lynch ☑️
@seanl It's not a Transparent Society if the citizenry doesn't have access to the surveillance and can't keep tabs on their government and law enforcement.

In conversation Tuesday, 08-May-2018 11:27:28 EDT from social.heldscal.la permalink
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 11:22:57 EDT clacke

This year Erlang celebrates 20 years as a Free Software programming language!

"""
1998 was an exciting year in which the following events occurred:
• The first demo of GPRS20 developed in Erlang was demon-
strated at the GSM World Congress in February and at CeBIT
in March.
• In February, Erlang was banned inside Ericsson Radio Systems.
• In March, the AXD301 was announced. This was possibly the
largest ever program in a functional language.
• In December, Open Source Erlang was released.
• In December, most of the group that created Erlang resigned
from Ericsson and started a new company called Bluetail AB.
"""

Joe Armstrong refers to this time as "Part IV: 1998 – 2001. Puberty problems — the turbulent years", which makes sense, as the language turned 12 in 1998. :-)

"A History of Erlang" (PDF) http://webcem01.cem.itesm.mx:8005/erlang/cd/downloads/hopl_erlang.pdf

In conversation Tuesday, 08-May-2018 11:22:57 EDT from social.heldscal.la permalink
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 08:39:50 EDT clacke
- Roland Häder
@roland It certainly failed fast when it didn't have enough memory.

After the install was done, most of that memory was still paged out, so this was not a case of using swap for the active set.

In conversation Tuesday, 08-May-2018 08:39:50 EDT from social.heldscal.la permalink
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 08:37:11 EDT clacke
- notklaatu
@cobra2 @klaatu There was no compilation involved. This was several GB of RAM burnt for the package system to figure out all the dependencies from the ground up. It might have gone differently if this weren't a relatively fresh install.

In conversation Tuesday, 08-May-2018 08:37:11 EDT from social.heldscal.la permalink
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 07:23:47 EDT clacke

Finally talked to the exchange. The reason my withdrawal isn't going through is because there was some KYC/AML thing sent last week that had got stuck in my spam folder.

I would have appreciated if the transaction status in the exchange had been shown as <awaiting customer response> rather than <processed>.

In conversation Tuesday, 08-May-2018 07:23:47 EDT from social.heldscal.la permalink
clacke (clacke@social.heldscal.la)'s status on Tuesday, 08-May-2018 06:56:58 EDT clacke
- Shellkr
@shellkr The size of /tmp or /dev/shm weren't an issue here, as increasing my swap could not have affected their size retroactively. Their sizes are set when they are mounted. The process simply ran out of memory, which I could also observe as the system ate all of the memory if had previously been restricted to, when it ran successfully.

Still, I usually set up my systems to have a lot of swap and an accordingly (explicitly) large /tmp, just because it's nice to have a big /tmp and I imagine without checking that tmpfs on swap is faster than a real filesystem, as tmpfs doesn't write to disk unless necessary, and when it does, it doesn't even try to have any kind of write barriers or other record-keeping for consistency between reboots.

Apparently building the kernel uses a lot of /dev/shm, maybe because of how gcc communicates with its own components? Nix doesn't communicate much, it only does pretty straight-forward process calls and in-process data processing.

In conversation Tuesday, 08-May-2018 06:56:58 EDT from social.heldscal.la permalink

After
Before

Public

Notices by clacke (clacke@social.heldscal.la), page 60