« first day (2952 days earlier)      last day (2003 days later) » 
06:00 - 16:0016:00 - 00:00

6:56 AM
 
7:30 AM
anybody knows what are these non printable characters symbols, please, I googled around but didn't find good result
<8C>
<85>
<86>
<87>
<88>
 
PLU NEL SSA ESA HTS
Unlikely that's what they're being used for, though
 
what are those?
where I can a reference table?
 
much appreciated
 
 
2 hours later…
9:22 AM
@Kusalananda Thanks.
 
9:58 AM
Rant: Spent some time trying to get a bioinformatics application working through Bioconda. After downloading and installing 2.5 Gb of stuff it failed because the "environment" that it sets up lacks the Module::Build Perl module. That's after installing Qt and at least one version of Python and Perl. At that point I'm just feeling so extremely old and old-fashioned.
I mean, I already have both Python and Perl on the machine, and I'll never use Qt for anything (not running X11 for starters). I absolutely hate virtual environments and I really don't get along with black box commands.
I'll just install the thing manually. Lot less hassle.
Also, I could probably run OpenBSD in 2.5 Gb of disk space. Why the heck conda needs to allocate that amount of space for every user that set it up is a mystery to me.
 
This is going to come across as entirely flippant, but I can run DOS in 2.5 MB of disk space ;-)
But yes, dependency management is bonkers nowadays (says the guy compiling Java with Maven all day long).
$ du -sh ~/.m2
12G	/home/skitt/.m2
:-/
 
10:19 AM
@StephenKitt Bonkers how?
@Kusalananda Distribution/OS?
 
@FaheemMitha there’s a big trend towards managing dependencies locally (inside each project), ignoring anything available on the system
and further along, rebuilding everything from source, and building everything statically
all these have their place but it’s nice when they’re not imposed on everyone
 
@StephenKitt Neither of those seem like good choices.
Though I'm not sure what a good choice would be. What Debian does seems to as good as anything, and better than most. But it still seems like making the best of a bad job.
 
10:37 AM
@FaheemMitha That was on Ubuntu Bionic.
 
@Kusalananda I see. Bionic, huh?
 
I suppose it's a way of ensuring a consistent environment, but still.
 
As far as I know, a good way of handling dependencies doesn't exist.
What Debian does is probably as good as anything, like I said. But it still has significant overhead and various problems.
Part of it might be that language ecoysystem rarely seem designed with a view towards dependency management.
Certainly the widely prevalent C model isn't.
 
@FaheemMitha Guix is arguably better, at least if you value reliability over disk space
 
@StephenKitt I'm not familiar with that.
 
10:43 AM
@FaheemMitha yes, and that’s been slowly changing; Go in particular has dependency management built-in
 
Meaning this, I suppose? en.wikipedia.org/wiki/GNU_Guix
@StephenKitt Oh. I didn't know that.
 
@StephenKitt Ok. Thank you.
 
 
2 hours later…
Tim
12:35 PM
@StephenKitt I was wondering which projects in particular are in that big trend?
Are dpkg and Guix used for dependency management of C and C++ libraries only?
@FaheemMitha What does deduplication mean in backup? Is it the same to diff which detects common lines and different lines between files?
-3
Q: What do comm and diff try to accomplish at input/output level?

TimGiven two files, for each line in each file, how do comm and diff determine whether the line also occur in the other file? if it does, whether its occurrences in the two files are the same or different? by taking into account the order between the lines in each file? How does diff decide...

 
@Tim it means that the same set of bits isn't stored in multiple places. Or to put it another way, there aren't multiple copies of the same set of bits.
You should be able to find stuff on the net that discusses it. It's relatively new, but I don't know the history.
For example, often people often have multiple copies of the same file in the same location. Frequently these files are very similar. They only differ in small ways.
Deduplicating backups will shrink the storage of these down to the space f one file plus the space required for the diffs for those files, roughly speaking.
Though I'm not familiar with the technical details.
 
Tim
12:52 PM
rsync can do deduplication too, right? If not, what feature of rsync am I mistaken for?
 
@Tim many of the “modern” programming languages with dependency management somewhat built-in (or used so much by its communities, in a unified manner, that it might as well be) exhibit this trend (Python, Ruby, Perl to some extent although it’s better integrated, Rust, Go...)
the main point of friction is that what makes sense for development doesn’t necessarily make sense for deployment, but the latter tends to be less of a concern for developers
@Tim no, there are Debian and Guix packages for the majority of programming languages in use currently
@Tim rsync can deduplicate during transfer, but not during storage — it doesn’t concern itself with storage, it relies on the underlying file system to take care of that
(so technically you can use rsync with storage deduplication, but it’s not handled by rsync)
many backup programs use their own storage formats and provide compression, encryption, deduplication, efficient storage of incremental backups etc.
 
@StephenKitt rsync can deduplicate? That's news to me.
 
@FaheemMitha well, not quite, it only transfers deltas
 
@StephenKitt That isn't the same as deduplication, is it?
 
@FaheemMitha no, not really, at least not storage deduplication
 
1:05 PM
If there are n identical files locally, presumably the remote rsynced version would also have n identical copies? And also I assume rsync would also transfer those contents n times.
 
@FaheemMitha not necessarily, because it compresses, so it shouldn’t transfer all the content as such
 
rsync is quite old. 1999? And just designed as an efficient transfer protocol.
@StephenKitt ok.
 
Tim
@StephenKitt Thanks. Another question is portability.

Kusalananda preferred restic over BorgBackup, because Borgbackup requires the target machine to also have borgbackup installed, while restic doesn't. Does rsync need the target machine to have rsync installed?

Is Borgbackup, Bareos, or restic implemented on top of rsync, or independently?
 
@FaheemMitha but I’m speculating somewhat, I’m not sure whether rsync compresses the stream or each file individually
it does have a nice feature where it can use non-transferred data in its compression
 
@StephenKitt Oh. How does that work?
 
1:08 PM
@Tim rsync works best when talking to an rsync dæmon, but it doesn’t need one
Borg, Bareos and Restic are independent of rsync
 
@FaheemMitha it compresses files it wants to transfer, locally (so it has the complete file information), but then only transfers the data corresponding to changes; the other end also has all the data required to reconstruct the result
 
> I couldn't help but ask, 'What if they block goes bad?' They said, 'Well that doesn't happen.'
Sounds like a reasonable question with a stupid answer.
The correct answer is, I suppose, do multiple backups preferably using different methods.
 
Tim
@StephenKitt @StephenKitt does it mean that Borg, Bareos and Restic are superior to rsync in all aspects, and can all replace rsync?
 
@Tim They are different from rsync.
Apples and oranges.
Borg and Restic are designed for backup. Rsync is designed for data transfer.
 
Tim
1:11 PM
what are common and different beween backup and data transfer
 
@FaheemMitha journalism as PR
@Tim no, they’re tools with different purposes
 
@StephenKitt PR?
 
@FaheemMitha public relations
that TechTarget article is mostly drinking the manufacturer’s koolaid
 
@StephenKitt Yes, I know.
 
Tim
@StephenKitt May I ask for details?
 
1:13 PM
@StephenKitt I don't follow.
In what fashion is it "drinking the manufacturer’s koolaid"?
 
Tim
@FaheemMitha they are advertising
taking money to promote the products.
 
@Tim Who is?
 
Tim
the host of the link you gave
 
@FaheemMitha it seems to me that the journalist talked to a couple of dedupe companies, and regurgitated what he was told without much critical thought
 
@StephenKitt That's certainly possible. I'm not saying it's a good article.
 
1:15 PM
@Tim data transfer takes care of “I have this data here, I want it there, make it so”
 
Tim
or they do not actually understand the products and features, but they think they do.
 
Just that it has a bit of a history overview. Though only of proprietary software, which is always more interesting. Though proprietary software people do often come up with new stuff first.
 
backup takes care of “I have this data, I want it preserved, make it so”
I belong to the Jean-Luc Picard school of specifications
 
Tim
@StephenKitt doesn't backup need to perform data transfer? Doesn't data transfer perform backup implicitly?
 
@StephenKitt In a nutshell. Apart from the Star Trek references.
 
1:17 PM
@Tim backup needs to perform data transfer, but data transfer isn’t backup
rsync will happily delete or overwrite data on the target
 
@Tim And you don't need to transfer everything. You could just transfer something.
Even just part of a file.
 
Tim
@StephenKitt rsync has an option referring to an old backup, and compare if the old backup already has some files that haven't been modified which it doesn't have to transfer. rsync also has options of what to do with a samename file on target: overwrite or not. Does that make rsync a backup tool beyond transfer one?
 
@Tim no, it means rsync can be used to build a backup tool, but it isn’t a backup tool on its own
 
@Tim It can be used as a primitive transfer too, or as the basis for a better one. But it's not a backup system per se.
 
Tim
Is it correct that rsync is both a data transfer tool and a backup tool?
 
1:22 PM
Well, at least its algorithms can be used to build a better backup too.
 
@Tim no
 
For example, the now largely dead rdiff-backup.
Which I believe used algorithms from rsync. I used to use it and liked it, but I don't believe it is deduplcating.
Hmm, Debian changelog last updated 04 Jan 2012. I'm actually a bit surprised it's still in Debian.
 
Tim
@StephenKitt Can you perform backup using a data transfer tool? Based on our discussion, I guess no?

Can you perform data transfer using a backup tool? If not a good idea, what of a backup tool do you don't want for data transfer?
 
Actually, last upstream release was 19 Mar 2009.
That's really a long time ago.
 
Tim
Is deduplication in backup the same as longest common subsequence problem? I guess no?
@MichaelHomer Hello, if I am correct, you are also a computer science guru. :)
 
1:29 PM
@Tim that’s the same as asking if you can hammer a nail using a chainsaw
technically you can, but it’s not a good idea
@Tim no, deduplication finds blocks (of a pre-determined, fixed size) which are identical, it’s not looking for the longest common subsequence
 
Tim
do you mean "can" to both directions? or just transfer by backup?
identical blocks means different blocks with identical contents, or the same block?
 
@Tim no, I mean technically you can hammer a nail with a chainsaw
@Tim why would you deduplicate the same block? it means blocks with the same content
 
Tim
@StephenKitt Thanks. do you mean both transfer by backup and backup by transfer, or just transfer by backup?
 
@Tim I mean both
 
 
1 hour later…
2:56 PM
@StephenKitt Have you tried restic? I think you wrote earlier that you use Borg.
And do you happen to run/maintain any mailservers?
 
@FaheemMitha I do indeed use Borg, and I haven’t tried Restic.
 
@StephenKitt Ok.
 
@FaheemMitha I used to but not for quite a while now (at least, not any Internet-facing mail servers).
 
Tim
May I ask what MUA, MTA or MDA are you using, Stephen
 
@StephenKitt Ok. I think it's gotten to be quite a hassle.
Russ Allbery once mentioned that he runs some. I wonder if he still does.
My email hosting provider has truly dreadful spam filtering, though otherwise it's satisfactory. I've been wondering whether trying to talk to them about it, or rather, complaining, would have any effect. Or whether I should look into switching hosting, which is a hassle, and has a non-zero chance of losing email.
 
3:00 PM
@FaheemMitha he still does AFAICT, dig -t mx eyrie.org
 
@StephenKitt Oh.
He also said he doesn't really recommend it. As an activity.
 
@Tim MUA: Claws Mail and Mutt; MTA: Exim and Postfix; MDA: procmail
with some lashings of fetchmail in some places
@FaheemMitha I agree whole-heartedly, it’s become too complicated.
 
Tim
What do you use MDA for? cron only uses MTA.
 
@StephenKitt Mostly spam issues? Or other stuff too?
 
@Tim do you know what all those tools do?
 
Tim
3:01 PM
somehow
 
@FaheemMitha stuff added to deal with spam ;-).
 
Spam is a huge pain in general. And those bastards keep evolving. They're like electronic viruses.
 
@Tim OK, so cron only needs to send email, so it only needs an MTA. MDAs distribute email, they’re used on the receiving end.
 
If there was a World Government, and they made spam a capital offense, it might have some effect.
 
Tim
How do you access gmail? Stephen
or mail yahoo?
 
3:03 PM
Neither.
 
Tim
Surprising...
oh, you use Office365?
 
hah
 
@Tim no
I use old-school SMTP/IMAP hosting with a hosting provider
 
Tim
Is your hosting provider free and available to public?
I am considering switching from Gmail
 
@Tim it’s not free, but anyone can buy hosting from them (OVH)
 
3:13 PM
@StephenKitt Any recommendations for email hosting? I've been using Luxsci since 2011 or so.
 
Tim
@StephenKitt May I ask what you don't like about gmail? (I know you don't have one, but if I may ask)
 
@FaheemMitha ProtonMail
 
@Tim You can get free providers, but you don't want to use them.
@StephenKitt Thank you. Is that what you use? And if so, for how long?
@Tim Google gets to go through all your email, of course.
 
@Tim I’ve had the same email address for 23 years, and I like knowing where the funding for the services I use comes from ;-)
 
Google also infamously cooperates with government agencies and will turn over all your emails without question
 
3:15 PM
@StephenKitt How is ProtonMail's spam filtering?
@derobert Any thoughts? And do you currently run any mail servers?
 
@FaheemMitha I don’t use it actively, I’ve had an account there since 2014
 
That being said everyone I've ever met with those concerns still owns a smart phone knowing that the mic is recording at all times and anything you say within earshot of it is being analyzed (likely by google)
 
@Jesse_b That's also true. But it's a relatively minor concern.
 
@FaheemMitha I don’t use it enough to say
 
HI @Jesse_b. Haven't see you around recently.
 
3:16 PM
@FaheemMitha Hola
 
@StephenKitt You wrote "I use old-school SMTP/IMAP hosting with a hosting provider". Were you referring to ProtonMail or something else?
 
Tim
@Jesse_b must have a baby
 
Yeah I've been in a funk
 
@Jesse_b I know that’s a popular theory, but bandwidth usage doesn’t really support it
 
And if you don't use ProtonMail much, who do you use instead?
 
3:17 PM
4 mins ago, by Stephen Kitt
@Tim it’s not free, but anyone can buy hosting from them (OVH)
 
@Jesse_b funk?
 
@StephenKitt It states right in the user agreements that everything can be recorded and is sent off to be analyzed
But it also perfectly explains why you simply speak about something and then you receive tons of ads for it
 
@Jesse_b can be
 
@StephenKitt Oh, right, sorry. But I'm confused. You're recommending a service you don't use much over one that you do?
 
Like 5 times now I have been on my way home from home depot or something and say to my wife "Oh crap I forgot to get deck screws"
 
3:18 PM
@FaheemMitha I don’t recommend most people do most of what I do
 
Get home and facebook has ads for deck screws on my timeline
 
with computers anyway
 
Tim
@Jesse_b does recording only happen when I talk over google voice?
 
@StephenKitt Google definitely does some processing on emails. When I send an email to someone with a Gmail account their name starts showing up in Linkedin, for example.
@StephenKitt How odd.
 
@Tim According to wikileaks the smart tvs record at all times
I would imagine phones are the same
 
3:19 PM
@Jesse_b You're kidding.
 
@FaheemMitha It happens all the time. Comedians make jokes about it now quite often
 
Tim
does it take screenshots all the times? when I type messages in google hangout, or whatsapp or slack?
 
@Jesse_b Good grief.
 
@FaheemMitha why? my use of computers isn’t typical
 
@StephenKitt Um. Well, a good hosting service is a good hosting service, surely.
So you don't recommend OVH, but do recommend ProtonMail?
And define typical.
 
3:20 PM
@FaheemMitha I recommend both, in different circumstances
 
@StephenKitt oh.
 
@FaheemMitha "I'm now getting ads for things I've simply thought about and never wrote or talked about out loud ever" is usually the gist of them
 
@Jesse_b I see. I don't recall seeing any such comedy routines. I suppose this is stand-up?
 
but for someone considering switching from GMail, ProtonMail is a better fit than an old-style hosting company
 
@Tim I'm not sure
@FaheemMitha Yeah I've seen it mentioned in at least one stand up act but I think it's normally like a twitter joke
 
3:22 PM
@StephenKitt Hmm. I'm not switching from Gmail, to be clear. Like I said, I'm using Luxsci. But their default spam filtering is pitiful.
 
I've had pretty good luck just using a throw away email
 
@FaheemMitha oh, sorry, I’m getting mixed up between the conversation with you and that with Tim
 
Possibly because they sell custom spam handling for an extra $2 a month. But I'm cheap. Plus I already pay them $10 a month for basic service.
 
I have an @live address that I use to sign up for things and that gets all the spam, my other addresses get virtually none
 
@Jesse_b What about posting online?
 
3:23 PM
@FaheemMitha What do you mean?
 
@Jesse_b Posting stuff online. Like, you know. words.
 
Almost all of my online accounts use my live address. I have a gmail that I give to friends and family
 
@FaheemMitha the thing I like about ProtonMail is that it enables encrypted email "for the masses" without having to mess around with GnuPG
 
@StephenKitt: I agree that cell bandwidth probably wouldn't even come close to supporting 24/7 streaming audio from all phones but thinking about it now it wouldn't be hard to just have the phone listen in for keywords and then send clips if deemed worthy. Similar to the way fraps can pre-record
I know I'm cynical but I just don't believe privacy is obtainable anymore. Short of going fully off the grid
 
@Jesse_b yeah that would be possible
or even just flagging, without sending the clips
because sending any amount of audio clips could still be spotted
 
3:29 PM
Yeah but if they are spotted the EULA says they can be there
 
@Jesse_b yes, at least in some jurisdictions ;-)
 
@Jesse_b Often your posts will be associated with an email address.
@StephenKitt That certainly is a nice feature. But I wonder how it handles the decryption part.
@Jesse_b Systematic use of encryption and refusal to use "free" services would go a long way.
I appreciate a bargain as much as the next person, but when people are giving away services for free, the correct response is "hang on a sec, why is (fill in the blank) being so nice"?
As I understand it, these days, all large companies are offenders.
@StephenKitt Overall, how do OVH and ProtonMail compare?
 
@FaheemMitha see
ProtonMail is an end-to-end encrypted email service founded in 2014 at the CERN research facility by Andy Yen, Jason Stockman, and Wei Sun. ProtonMail uses client-side encryption to protect email contents and user data before they are sent to ProtonMail servers, unlike other common email providers such as Gmail and Outlook.com. The service can be accessed through a webmail client, the Tor network, or dedicated iOS and Android apps.ProtonMail is run by Proton Technologies AG, a company based in the Canton of Geneva, and its servers are located at two locations in Switzerland, outside of US and EU...
 
@FaheemMitha Yeah, many large companies make a significant portion of their income by selling customer information
 
@FaheemMitha ProtonMail only does email and VPN, and is pretty good at least at the former
OVH is a wide-spectrum hosting company
 
3:40 PM
@StephenKitt I see. But you don't know how their spam handling is?
 
If you sign up for a "savings" card with your grocery store they are selling your name and email address
 
they’re pretty good at most of what they do
but not excellent at anything in particular
@FaheemMitha no, nor do I know how OVH’s spam handling is
I do my own
 
I don't understand why Luxsci spam handling is so bad. I should create a ticket. You know - what the hell?
 
@Jesse_b again, that depends on the jurisdiction
 
@StephenKitt Oh, you do? Ok.
@Jesse_b As I understand it, that's the point of those cards. They lure you in with the discounts. Possibly it also helps them keep track of your purchases.
 
3:42 PM
@terdon I put up a flag, in case you catch it, and in case you want to ping me about it.
 
@FaheemMitha Yeah so they can target advertisements to you based on your spending habits
 
I mean, if I was them, and, you know, evil, I'd be tracking purchases too.
@Jesse_b Exactly. Do you find you get more ads in your mail for things you buy regularly with your card?
 
Developing advertisement algorithms is a huge business now pioneered by companies like facebook and google
@FaheemMitha I do but they also give me coupons for those things and I'm not that worried about it
 
@Jesse_b I should get a job doing that. Then I can put Evil Genius on my business carrds.
 
@Jesse_b that’s really the main point of grocery store cards (and any such cards in other types of stores)
 
3:44 PM
@Jesse_b I don't think that particular issue is worth being worried about so much.
 
at least for major grocery chains, the name of the game is to maximise spending, and they do that by encouraging you to buy stuff you normally buy
 
@StephenKitt What's the main point? To give their customers coupons, or something else?
 
it’s a feedback loop
 
Yeah it was just an example
 
large grocery chains are banks effectively
 
3:45 PM
@StephenKitt A limited feedback loop, though.
 
but what you buy at one store is valuable information for other advertisers and that information does get sold
 
@Jesse_b again, that depends on the jurisdiction
 
I mean, there is a limit how much more people will buy, coupons or not.
 
@StephenKitt I'm sure but I think in the US there are no limitations on that sort of thing
 
@FaheemMitha yes, but store A wants to ensure you buy what you do buy in store A rather than store B
 
3:45 PM
and even if there are, there is certainly some loophole around it
 
@Jesse_b probably not
 
we have loopholes for everything
 
I used to make a point of ignoring coupons when I lived in the US. I didn't like having my spending habits manipulated. Plus, I don't particularly recall them sending me coupons for stuff I purchased.
@StephenKitt Good point.
Though that could turn into an arms race. But I suppose it doesn't in practice because not enough people make large purchases at multiple stores.
 
@JeffSchaller Heh, I just saw this now while coming here to ask you about it :)
 
Anyway, how have you been @FaheemMitha? Still having issues with your building management?
 
3:53 PM
@Jesse_b Oh, that. Yes.
But currently I've got worse and more pressing things to worry about. But thanks for asking.
 
Sorry to hear
 
Oh, and to answer your questions, things have been fairly bad. Partly my own fault, I suppose. I've not good enough at watching my back.
@Jesse_b That makes two of us.
 
Anything you want to talk about?
 
@Jesse_b Sure. But I'm trying to finish typing up something right now. If you are around, maybe later.
So, did you have a baby? Or did I get you confused with someone else?
 
Don't let it affect you though. How many times you fall down is irrelevant, the only thing that matters is how many times you get back up (cliche :p)
Not yet, he's due next month
 
3:58 PM
@Jesse_b But making life busier, I bet.
 
@FaheemMitha I still run my own mail server, with exim + dovecot. Spam filtering is done via greylisting & spamassassin
 
@Jesse_b Well, in bad situations, you have to keep your cool. Otherwise you'll have a heart attack, or a nervous breakdown. But that's not really the same thing as you said.
@derobert So no email hosting? Was looking for recommendations. Spam filtering isn't particularly easy, but I also don't think it's hard to do a reasonable job. Which makes me wonder why Luxsci is so pitiful at it.
 
06:00 - 16:0016:00 - 00:00

« first day (2952 days earlier)      last day (2003 days later) »