/dev/chat

Paṇḍyā

1:27 AM

Just achieved 10k on U&L :)

Kusalananda

6:21 AM

@Paṇḍyā Congratulations!

Paṇḍyā

7:52 AM

@Kusalananda Thanks.

terdon

9:22 AM

@Paṇḍyā Nice! Well done! :)

TheLittleNaruto

10:12 AM

@Paṇḍyā Congrats! :)

Jeff Schaller

11:03 AM

@Paṇḍyā congratulations!

Paṇḍyā

12:55 PM

@terdon :)

Thanks to all.

Jesse_b

3:57 PM

foo=(\<\<\<Hello\\ chat \> /dev/chat cat)
eval "${foo[@]}"

Sometimes I feel like I'm the only one looking for new and more obnoxious ways to write hello world

Jeff Schaller

@Jesse_b wall Hello Jesse

Jesse_b

@JeffSchaller =)

Stephen Kitt

@JeffSchaller I wondered where that message on all my terminals came from

Jesse_b

Hey @JeffSchaller: It looks like the whole post may have been deleted but earlier there was a post where a user was asking about encrypting a shell script. I made a comment that I didn't understand their question and they then edited their question to simply be a sarcastic reply to my comment

Stephen Kitt

@Jesse_b read the 51st character of all my answers here, in the order I wrote them (just kidding)

Jesse_b

4:08 PM

I then reverted their post back to the original question. Is that the right move?

@StephenKitt Hah that would be pretty epic if you really did something like that

Jeff Schaller

@Jesse_b you could try it once, but in the heat of the moment (and if they can find the edit/rollback buttons), they may just get into an edit/rollback war. I'd cut my losses and VTC & move on.

@StephenKitt

wall we would like to apologize for the previous obnoxious message with this obnoxious message and promise to do better with our obnoxious messages in the future

2

Jesse_b

@JeffSchaller Cool thanks, yeah I wasn't trying to get into an edit war but I know I was once told to not edit spam answers as it would make them less likely to be downvoted and wasn't sure the stance on questions

Stephen Kitt

@JeffSchaller you forgot a few Ctrl+Gs in there if you really want to be obnoxious ;-)

Jeff Schaller

@Jesse_b if they've edited into something flag-worthy (spam or abusive), then leave it & flag as such

@StephenKitt leaving room for improvememnopqrst

lol

Jesse_b

4:31 PM

Holy crap. I need to convert a given date stamp to epoch seconds in solaris and apparently there isn't really a way to do that in solaris

Found a forum post where they suggest reinventing the wheel

Faheem Mitha

@Jesse_b Does your Solaris installation have GNU tools?

Jesse_b

@FaheemMitha It doesn't

Jeff Schaller

@Jesse_b step 1. find some mud and straw. 2. make bricks. step 3. find sticks and make fire. step 4. install Linux

@Jesse_b perl ?

Jesse_b

@JeffSchaller Yeah I may have to use perl. I found a few possible solutions in node as well

I'm actually sort of interesting in trying this "code" out though

if month > 2 then
  month=month+1
else
  month=month+13
  year=year-1
fi

day=(year*365)+(year/4)-(year/100)+(year/400)+(month*306001/10000)+day
days_since_epoch=day-719591 (which is Jan 1 1970)
seconds_since_epoch=(days_since_epoch*86400)+(hour*3600)+(minute*60)+seconds

Obviously with improvements to make it actually run, but I wonder if that formula is solid

Faheem Mitha

@Jesse_b Could you install them?

Jesse_b

4:34 PM

@FaheemMitha I cannot

Faheem Mitha

Ok.

Jeff Schaller

4:50 PM

@Jesse_b perl -MPOSIX -e 'print strftime("%s\n", 0,0,0,1,0,90)' -- for Jan 1 1990; year -= 1900; month is zero-based

derobert

@JeffSchaller unix.stackexchange.com/posts/548348/revisions ... appears OP gave up and deleted it.

Jeff Schaller

@derobert yes, thanks to Jesse_b 💘💘

I missed that part while closing it

I might have saved a life with an answer, but Jesse_b has found true love

Jesse_b

@JeffSchaller Well if people are just tossing doable shell scripts around, I'm going to grab one up for myself

derobert

@Jesse_b Hmmm. I can figure where most of that came from (e.g., the 4/100/400 is the Gregorian leap year rules)... The 306001/10000 is interesting; maybe that's average month length? Does it work? Who knows...

Jeff Schaller

I'm also getting 18000 for Jan 1 1970, not 719591. 719591 seems to be Jan 9th 1970

derobert

4:56 PM

@Jesse_b actually, no, it doesn't work. There is no way that month adjustment can. February is evil.

Jeff Schaller

oh that's days. nevermind

derobert

e.g., March 1 should fail. I'll be shocked if it doesn't.

Jeff Schaller

no, I still don't get it. oh well. I also vote "do not use this code"

derobert

@JeffSchaller 1970*365+1970/4-1970/100-1970/400 = 719519

(all of that done with integer math, i.e., just paste it into bc)

Jeff Schaller

oh so that's seconds since the century?

I'm not smart enough for this

derobert

5:01 PM

No, that should be .... well, an arbitrary number, really. Days since year 0 (if that existed) on the Gregorian calendar (which doesn't go back that far). But if you subtract the calculation of some sane date (say, jan 1 1970) from it, it'd be days since that date

Except it seems the adjustment is 72 days off...

Jeff Schaller

to paraphrase Click & Clack: "booooooogus!"

derobert

Oh, let me try not assuming 0-based day/month...

(1970*365)+(1970/4)-(1970/100)-(1970/400)+(1*306001/10000)+1-719591
-41

.... so, no, that didn't fix it. Just seems broken. Oh wait, that if in front must be part of it..

(1969*365)+(1969/4)-(1969/100)-(1969/400)+(14*306001/10000)+1-719591
-8

closer!

ilkkachu

@Jesse_b where's that code from?

derobert

If you want a formula, I bet the one under "Converting Gregorian calendar date to Julian Day Number" at en.wikipedia.org/wiki/Julian_day works....

That will get you a day number. You then subtract the day number of Jan 1 1970 from it. Then you have a number of days, which is easy enough to convert to seconds on the pretend-there-are-no-leapseconds scale POSIX uses (multiply by 86400). Hours, minutes, and seconds are easily converted to seconds as well, and so is the timezone offset (since epoch timestamps are always UTC)

Or, if you're sane, just use perl.

Jesse_b

@ilkkachu unix.com/shell-programming-and-scripting/…

ilkkachu

5:15 PM

cute

Jesse_b

wheelv2 () {
	local d=$1
	local _year _month _day _hour _minute _second epoch day days_since_epoch seconds_since_epoch
	local pattern='[0-9][0-9][0-9][0-9]-[0-9][0-9]-[0-9][0-9]T[0-9][0-9]:[0-9][0-9]:[0-9][0-9]Z'
	if ! [[ "$d" =~ $pattern ]]; then
		printf '%s\n' 'Error: Unknown input date format' >&2
		exit 1
	fi
	_year=${d:0:4}
	_month=${d:5:2}
	_day=${d:8:2}
	_hour=${d:11:2}
	_minute=${d:14:2}
	_second=${d:17:2}
	if [[ "$_month" -lt 2 ]]; then
		_month=$((_month++))
	else
		_month=$((_month+13))

$ ./reinvented_wheel.sh '2019-10-23T16:28:43Z'
Input date is: 2019-10-23T16:28:43Z
Your results are as follows
GNU Date: 1571848123
Reinvented wheel: 1572020923

ilkkachu

don't you have that condition wrong? -lt instead of -gt?

Jesse_b

oh

good catch

ilkkachu

there's some complete shell script of that here: github.com/Apress/exp-shell-scripting/blob/master/Chapter_03/…

Jesse_b

Still not accurate

ilkkachu

5:23 PM

that script I linked gave me a number that was off by 10800 seconds. Which is exactly the difference to UTC here right now... Using date -u would fix that.

so yeah, that script seems to work

the book was on Springer, so I took a look, but it didn't give a full script, and in particular, didn't tell where that 306001/10000 came from

Jesse_b

I'm inputting a UTC time stamp and it's still off for me

Jeff Schaller

@Jesse_b are you comparing to GNU date? and are you using date -u ...?

ilkkachu

but putting that equation for 1970-01-01 into bc, it gives me 719591.883900, so I'm pretty sure there's some curve-fitting and creative use of integer rounding involved

Jesse_b

@JeffSchaller gist.github.com/jessebutryn/c7e30f20ff05da10444639f2dcc3bf6d

Jeff Schaller

@Jesse_b maybe you want gdate -d "$_date" '+%s -u' ?

Jesse_b

5:27 PM

That would print a literal -u, no?

ilkkachu

@Jesse_b, _month=$((_month++))!!!

Jeff Schaller

I might be confused with ilkkachu's comment regarding a timezone's worth of error

ilkkachu

If this only was C code, the compiler would scream at you

Jesse_b

The epoch seconds should always be in UTC though right?

ilkkachu

$ _month=123; _month=$((_month++)); echo $_month
123

Jesse_b

5:29 PM

@ilkkachu aha!

It does seem to work now :O

ilkkachu

x++ returns the old value and increments after. ++x increments first and returns the new value.

@JeffSchaller yeah, the script I linked had date +%k, which gives the hours in local time. So it's off by the time zone when compared to date +%s.

derobert

Does it work for 1970-01-01T00:00:00Z ?

Quickly running it here (well, or rather on some random web codepad I found), gave 31536000 ... which is quite wrong...

Wait, that year-- should be year-1

ilkkachu

works

derobert

Then it appears to work...

Jesse_b

so

I've been told to not use perl since it shouldn't be relied on to always be available

So that wheelv2 shell function might make it into some production code here shortly lol

ilkkachu

5:41 PM

hmm

argh, timegm() isn't standard either

Jeff Schaller

@Jesse_b can't think of a Solaris (or even SunOS) system that didn't have perl; bummer

ilkkachu

and mktime() assumes the input is in local time. I would have suggested making a short C program to call one of those, but. yeah.

Wildcard

I can see how Go's approach to dependencies can be so attractive. Just compile them all into the same binary so you don't have to think about it.

Jesse_b

@JeffSchaller We definitely have perl but we have removed the version that ships with the OS and now we only have a secondary install. The fear is that secondary install will also eventually go away

So FYI, current versions of smartOS don't ship with perl

Jeff Schaller

@Jesse_b and they say there's no such thing as a timeless system

Jesse_b

5:50 PM

We run the OS off a ramdisk that is like 256MB or something and they don't want to increase the size of our ramdisks so instead they have been looking for ways to shrink the OS

removing perl was one of those ways

Jeff Schaller

that's gotta save at least 300 Mb ;)

Wildcard

@Jesse_b I wonder at what point the removal of perl would be outweighed by the length of scripts that have to work around the absence of perl....

Jeff Schaller

random RPM of perl5 shows ~23 Mb, so 10% is decent

Jesse_b

@Wildcard Apparently when they looked into the feasibility of it there were only a handful of things still using perl and many of them were operations tools that were barely used

Jeff Schaller

silly things like converting dates correctly

ilkkachu

5:54 PM

@JeffSchaller uncompressed?

Jeff Schaller

sorry, @Jesse_b; giving you grief is more fun than setting up this printer driver

@ilkkachu Size : 23552149 and du -sm /usr/share/perl5 gives 23

Wildcard

I suppose using the dateutils package is out of the question?

fresse.org/dateutils

Jesse_b

@JeffSchaller Heh, no worries. I appreciate it because they are all valid points, unfortunately logic doesn't always win I think

Wildcard

Maybe just set up a microservice so you can use a REST API to convert all your dates. ;)

Jeff Schaller

@Wildcard in a container

in a zone

ilkkachu

5:56 PM

yeah, I suppose it would be rather big if that was the compressed size...

Wildcard

@JeffSchaller with auto-scaling.

Jeff Schaller

in the cloud

it should be multi-tenant, and if you're sure that the timezone issues are worked out, you can make it multi-region

Wildcard

I think the npm fiasco about leftpad wasn't nearly as silly as we could get if we tried. How about making leftpad an auto-scaling microservice instead of a package?

Jeff Schaller

> Have We Forgotten How To Program?

I missed the whole fiasco, for some version of "missed"

Wildcard

@JeffSchaller yep, that's the one.

ilkkachu

6:00 PM

I'd heard about leftpad, but this: "There’s a package called is-positive-integer (GitHub) that is 4 lines long and as of yesterday required 3 dependencies to use." !?

Wildcard

But really, my answer is: no, "we" haven't forgotten how to program.

ilkkachu

how does one even

Wildcard

Some people never knew how in the first place.

I think it's the result of programming becoming "cool" and a highly paid job.

Before that, only people who were actually interested in computers learned to program.

So there are now a LOT of people "working as programmers" who don't have a clue.

@JeffSchaller well, I did too, sort of. I heard about it at the time, but I never dealt with npm directly.

@ilkkachu they should have made it a microservice to reduce dependencies. :P

ilkkachu

I suppose one would need to have some clue to build something as high-profile as React

actually that dependency on left-pad somehow makes me think that there's a lack of a proper standard library which would include the basic stuff like that

Faheem Mitha

Looks like today's Debian's google-chrome upgrade might have broken widevine on Chromium, since they moved the location of the embedded widevine so. This affects me on Amazon Prime.

Can anyone else reproduce?

Wildcard

6:04 PM

@ilkkachu really it comes down to archive.ece.cmu.edu/~ganger/712.fall02/papers/p761-thompson.pdf

Except that it's trusting competence, not trusting trust.

ilkkachu

because, sure, you can implement padding on your own, but calling an existing function is one line, while that implementation in the article is ten

Faheem Mitha

The good news is that using the new location appears to fix the issue. At least till such time as Google has another bright idea.

Jeff Schaller

@Wildcard that paper is amazing; blew my mind the first time I read it, and every time after that

Faheem Mitha

It's actually surprising that the widevine so works at all. I would expect ABI incompatibilities.

Jeff Schaller

> The moral is obvious. You can't trust code that you did not totally create yourself. (Especially code from companies that employ people like me.)

Faheem Mitha

6:06 PM

@JeffSchaller You can't trust code you wrote yourself, either. Because it has bugs in it.

Solution: move to the woods, build a hut, and meditate on nature.

ilkkachu

take up growing cabbage and herding goats

Faheem Mitha

Optional extra step: write a book about yourself and become famous.

I recently wrote some Lua backend code for use with LaTeX (LuaLaTeX). It's actually working, and it feels kind of weird.

ilkkachu

the goats might be less stressful than computers but fame might quickly undo that

Faheem Mitha

I even got it working against a fairly short deadline, the other day.

ilkkachu

@Wildcard err, I'm not exactly sure how that works with competence?

Faheem Mitha

6:09 PM

@ilkkachu The really important question is whether goats are more or less stressful than TeX. At least they don't produce incomprehensible error messages.

Wildcard

@ilkkachu Well, you wrote:

> I suppose one would need to have some clue to build something as high-profile as React

And I'm pointing out that even if you do have a clue, you are building on the work that came before you.

Faheem Mitha

I don't know how well other people program. On all the projects I ever worked on with people, they refused to. With surprising reliability, actually.

Wildcard

You're trusting the competence of the chip designers, the OS designers, etc., etc.

Faheem Mitha

Which is kind of frustrating, actually. Because I like to learn from other people.

Wildcard

And in the npm ecosystem, that means you're trusting other developers who randomly trust other developers who trust other developers, without having any really good reason to do so and without doing any code review.

ilkkachu

6:11 PM

@Wildcard ah, yes yes, of course.

Faheem Mitha

BTW, if you are in the mood for a semi-horror film, check out Escape Room.

It's much better than you would expect.

Wildcard

I read an article or a book chapter on the subject one time, I think in the context of why Google moved most of their dependencies in house rather than assuming that random outside people are competent.

ilkkachu

something threw me on a really off-track here. May have been the reference to that paper.

Faheem Mitha

@Wildcard If you think about it, it's surprising everyone's computer doesn't just explode.

Wildcard

@FaheemMitha yeah, except in the realm of security.

(Where they do explode, pretty much.)

Faheem Mitha

6:13 PM

Google's approach to packaging standards is to ignore them. Which is nice. Look at their Debian changelogs. It's the current package.

Also, they install in opt/.

Which is just insulting.

@Wildcard Metaphorically, yes.

@Wildcard Does that mean they are assuming that their random in-house developers are competent?

Wildcard

@FaheemMitha my comment is really a very bad mischaracterization of the article.

It was extremely thoughtfully written and applicable to any team.

ilkkachu

@FaheemMitha I suppose it's also relevant to be able to apply in-house processes to the code

Wildcard

It's just about the tradeoffs involved in using external dependencies, and what you should think through before you accept a new dependency into your project. i.e., evaluate the quality of the code, how well maintained it is, etc.

Faheem Mitha

Doesn't Google have like have tens of thousands of employees now?

I wonder how many of them program.

ilkkachu

"Number of employees 103,459 (Q1 2019)" (Wikipedia)

Wildcard

6:19 PM

@FaheemMitha I can't make sense of this comment.

Faheem Mitha

@Wildcard They just include the most recent version.

Except they don't change the date.

Wildcard

@FaheemMitha who includes the most recent version of what?

Faheem Mitha

So the version that was installed today is.

google-chrome-stable (78.0.3904.70-1) stable; urgency=low

  * Build spec: chromium.googlesource.com/chromium/src.git/+/78.0.3904.70
  * Release Notes:
    googlechromereleases.blogspot.com/search/label/Stable%20updates

 -- Chrome Linux Team <[email protected]>  Tue, 03 Feb 2009 14:54:35 -0800

@Wildcard Google (presumably) only includes the most recent version of Google Chrome in their Debian changelog.

Note the 2009. It's not 2009.

Wildcard

@FaheemMitha maybe they didn't have perl available to convert the date.

Faheem Mitha

@Wildcard ?

Wildcard

6:22 PM

@FaheemMitha it's tongue in cheek.

ilkkachu

@FaheemMitha I suppose you missed the previous conversation...

Wildcard

Start from here: chat.stackexchange.com/transcript/message/52262213#52262213

Faheem Mitha

@ilkkachu Something about Jesse trying to convert something?

Some time related thing to Unix seconds?

ilkkachu

@FaheemMitha which would have been easy with Perl, but the system they were working on didn't have it. they'd removed it to save space.

Faheem Mitha

@ilkkachu That's aggravating. Perhaps they should get more space.

Wildcard

6:28 PM

37 mins ago, by Jesse_b

We run the OS off a ramdisk that is like 256MB or something and they don't want to increase the size of our ramdisks so instead they have been looking for ways to shrink the OS

Faheem Mitha

@Wildcard I wonder why they don't want to increase the size.

Wildcard

@FaheemMitha to improve security, probably. :P

Faheem Mitha

@Wildcard Does that really improve security?

Wildcard

@FaheemMitha yes, because there is less space in which viruses could be present, right?

Faheem Mitha

@Wildcard Ummm...

Wildcard

6:30 PM

LOL

Jesse_b

In this case I think we mostly just wanted to avoid using more memory unless absolutely necessary

The engineering team seemed to think it was feasible to remove perl which it must have been because they did it :p

Faheem Mitha

That wouldn't fly on Debian.

Jeff Schaller

@FaheemMitha would it crawl?

Faheem Mitha

@JeffSchaller Debian doesn't crawl. It soars.

Wildcard

@FaheemMitha looks like the Chromium changelog is just the git log.

chromium.org/developers/change-logs

Jeff Schaller

6:38 PM

> There is no greater joy than soaring high on the wings of your dreams, except maybe the joy of watching a dreamer who has nowhere to land but in the ocean of reality.

@JeffSchaller (despair.com)

alright, I won't quit my day job. back to printers.

Wildcard

@FaheemMitha do you work much with TeX?

Faheem Mitha

@Wildcard It's customary to leave the earlier entries.

@Wildcard A fair amount. Why? I do all my writing in it. They like forms in India. And traditional letters.

I think they're moving away from that in the West.

Wildcard

@FaheemMitha I poked into it a very tiny bit.

Faheem Mitha

@Wildcard It's handy if you want to write stuff, and you care about typography at all.

Wildcard

I wanted to see if there would be a decent template I could use to reconstruct the particular look of a checklist that is currently in InDesign.

I couldn't find anything.

Faheem Mitha

6:43 PM

TeX tends to be rather DIY.

Wildcard

I was looking into LaTeX, actually.

Still didn't find anything. Maybe I didn't look long enough but that's why I'm asking.

Faheem Mitha

The TeX SE site and chat are very helpful.

@Wildcard Well, yes. LaTeX is TeX.

It's just TeX macros. It's all TeX under the hood. It's just hidden.

Wildcard

@FaheemMitha yeah, but slightly less DIY, at least there are more packages and features available.

@FaheemMitha yes yes, of course. :)

Faheem Mitha

@Wildcard Nearly everyone uses LaTeX. But it's still better to think of it as TeX. Because that's what it is.

@Wildcard You could ask on TeX SE. They're fairly nice to beginners.

At least compared to places like SO.

They do kind of expect you to try first, though.

How much have you used TeX?

Wildcard

@FaheemMitha hardly at all.

I think I will ask a question.

What I ran into is that there is no document type that looks appropriate for what I want.

Faheem Mitha

6:49 PM

@Wildcard Maybe run through a tutorial that covers the basics? Otherwise you might not understand the answer.

TeX isn't that hard, really, but it has a learning curve. And it doesn't help that there is really nothing else like it.

Wildcard

@FaheemMitha I understand the basic formatting commands. Anyway, I'll write up a question.

First I need to have lunch. :)

Faheem Mitha

@Wildcard Ok.

Jesse_b

7:33 PM

#!/usr/bin/env bash

genr () {
	printf '%s %s' \
		$(cat /dev/urandom | gtr -dc 'a-z' | fold -w 5 | head -n 1) \
		$(cat /dev/urandom | gtr -dc 'a-z' | fold -w 5 | head -n 1)
}

until [[ "$(genr)" == 'hello world' ]]; do
	:
done

echo "Took $SECONDS to generate hello world"

Jeff Schaller

@Jesse_b go for the glory and allow a space and require hello world all in one go!

Jesse_b

@JeffSchaller I still haven't gotten it to finish at all lol

I did get this on like the third test I ran with the random command:

$ cat /dev/urandom | gtr -dc 'a-z' | fold -w 5 | head -n 1
oorld

so close to world lol

Jeff Schaller

testing to see if the CPU alerts are working? :)

Jesse_b

@JeffSchaller Whenever I have to write code that I don't want to write I follow it up by writing something completely pointless that interests me

I wonder what the odds of it hitting hello world even are, would it be 10^26?

Jeff Schaller

@Jesse_b balance in the universe, I like it

@Jesse_b time for another shell script!

Jesse_b

7:39 PM

:p

it's actually only using 5% of my CPU at the moment, anti-virus scan is using 30%

ah I think it would be 26^10

Jeff Schaller

well, if I was an antivirus program and I saw some user piping two /dev/random's to three other commands, I'd be a little concerned, too

Jesse_b

@JeffSchaller Well if you were an antivirus program, the antivirus program would become Jeff Schaller and it would use your body to scan my PC. You can't stop me no matter who you are!

Jeff Schaller

then, apart from deleting some nonsense on U&L, I will have to chalk today up to mostly-failure

Jesse_b

@JeffSchaller It's a good thing the antivirus program doesn't have control of your body. They love to delete things

Jeff Schaller

@Jesse_b thinking of bodies and antivirus reminds me of a recent podcast I heard about CRISPR... which one was it

gastropod.com/whats-crispr-doing-in-our-food I think

pretty amazing stuff

Jesse_b

7:53 PM

decent

super GMO?

Jeff Schaller

good GMO, bad GMO? I'm just impressed at what science can do

should we mess with nature? too late, we have. but it's interesting

Jesse_b

Well I'm certainly not against GMO

I've never met anyone against it that even knows what that means

They still eat bananas and corn

Jeff Schaller

bingo

Jesse_b

My computer is still searching for hello world

if I did my math right which I absolutely may not have, it should take at most 11293367652270 seconds

That's only 358110 years

derobert

@Jesse_b (1/26)¹⁰ or 1/(26¹⁰).

Jesse_b

8:03 PM

seems like it should be faster than that, or I can start using shorter passwords

derobert

Your chance of hitting it in the first N attempts, though, of course grows. But not the nice linear way you'd expect (because, e.g., it's possible to guess the same thing twice, since its random)

1-(1-(1/26^10))^n I'd guess...

ilkkachu

@Jesse_b urandom is slow, and filtering with tr -dc a-z throws around 90 % of the bytes away

derobert

Wolfram Alpha tells me you'd need something around the order of 87 trillion attempts to have a 50% chance. At least, if my guess at the formula is correct.

Jesse_b

I'm trying just hello now

until [[ "$(cat /dev/urandom | gtr -dc 'a-z' | fold -w 5 | head -n 1)" == 'hello' ]]; do
	:
done

Wildcard

@FaheemMitha I asked it.

0

Q: Is there a LaTeX document type appropriate for a multi-page, multi-section checklist?

I am a beginner to TeX and LaTeX. (Though I have used Softcover a little bit, which uses LaTeX under the hood and accepts some LaTeX commands, and I am very experienced with multiple flavors of markdown.) I have a multi-page, multi-section checklist that is currently very nicely formatted in In...

document-classes

ilkkachu

8:15 PM

actually, that cat | tr | head could throw away even more of what is read from urandom if cat has enough time to fill the pipe buffer before it dies from the pipe breaking

even if it reads a block of just 4096 bytes and passes it on the pipe, there'd still be on average something like 400 lowercase letters, of which you're only using five

Faheem Mitha

@Wildcard Ok. You'd normally get a fast reply, if you get a reply at all. Under 24 hours, usually.

The people who are answer questions are relatively few. It's mostly the same few people who answer my questions.

Often the top two people by rep - David Carlisle, and egreg.

Jesse_b

@ilkkachu this seems significantly faster:

head -n1 /dev/urandom | gtr -dc 'a-z' | fold -w 5 | head -n 1

Faheem Mitha

@Wildcard They're going to want something more concrete than that. At least an example of what you are talking about.

I suggest uploading a PDF of the thing. Though I can't remember if SE allows PDF uploads.

And ideally an example of an attempt to reproduce it.

ilkkachu

@Jesse_b uhm, yeah. on my Debian, cat reads 128 kB at a time, head just 8192 bytes. A 16x difference

derobert

@Jesse_b that also doesn't work

Faheem Mitha

8:20 PM

@Wildcard take a look at enumitem, to start with.

derobert

(there is a 1/256 chance of each character being LF, which would stop the head early)

Faheem Mitha

That sounds like it's in the ballpark, at least.

Wildcard

@FaheemMitha yeah, problem is, the particular document I'm working with is not publicly available and is definitely not okay to put in Creative Commons.

ilkkachu

now, what are the odds that there would be less than five lowercase letters before the first LF?

Jesse_b

@derobert Well considering it's over 8x faster I think even if it fails 1 in 256 times it's still faster overall

Wildcard

8:20 PM

So I would have to make a dummy document just for upload.

Faheem Mitha

@Wildcard Generate a similar and small example, and upload that.

texdoc enumitem

You can do pretty much anything with TeX. These days TikZ even allows animation. Which I have mixed feelings about.

derobert

@Jesse_b Could also try have gtr read the file. Just redirect < /dev/urandom

ilkkachu

...something on the order of 18 % (to fail), it seems

Jesse_b

yeah

$ for ((i=1;i<=100;i++)); do var=$(head -n1 /dev/urandom | gtr -dc 'a-z' | fold -w 5 | head -n 1); [[ ${#var} -ne 5 ]] && printf '%s\r' "Failed $((++fail)) times"; done; echo
Failed 13 times

ilkkachu

(26/27)^5 would be 82.8 %, if that's even the correct expression

Faheem Mitha

8:25 PM

@Wildcard If David replies, he'll complain that your question is unclear.

At least, he always does with mine.

Jesse_b

Faheem Mitha

@Wildcard Presumably you are aware that document classes are extensively customizable? The KOMAScript classes go particularly far in that direction.

But what you are describing doesn't sound particularly exotic.

ilkkachu

@Jesse_b on my system, tr manages to read a couple of blocks from urandom before dying on the pipe, while head -n1 stops after the first one (with a high probability)

Faheem Mitha

@derobert Some finance magazine published a list of the world's 50 safest banks. Spoiler: none of them are Indian. There's also a list of the top 10 for Asia. India isn't in that one either.

derobert

android.stackexchange.com/questions/217904/… ← somehow avoided titling that question "How do I teach Google to count?"

ilkkachu

8:31 PM

so, if it's indeed the speed of urandom that's throttling you, that would affect it

Wildcard

@FaheemMitha I dunno, I think my question is pretty explicit:

> So my question is: Is there a document type that is appropriate for this type of checklist?

And a secondary question: If there isn't such a document type that's a perfect fit, what are the tradeoffs I will have to deal with in using various other document types to produce this checklist, i.e., which document type fits best?

(Or would I actually be better off writing my own document class, even though that seems like a tremendous hurdle?)

@FaheemMitha there may be something fundamental I'm missing.

Faheem Mitha

@Wildcard Clearly you've not spent much time around TeX people. :-)

Wildcard

Like, maybe it doesn't actually matter what document type I use, it only matters what package I'm using.

Faheem Mitha

Seriously, just upload an example.

Wildcard

But all the tutorials I've seen tended to imply otherwise.

Faheem Mitha

8:33 PM

It would take you like 20 min.

Wildcard

Yeah...the trick is, to make the example I would have to pay for InDesign again.

Faheem Mitha

@Wildcard The document classes don't do that much, in general.

Just some basic stuff.

@Wildcard Huh?

You don't have the software installed?

Wildcard

Right now, we only pay for the one month subscription once in a long while, when we absolutely need to update the checklist.

That's part of the reason it would be nice to transition, so that we could more easily make small changes without needing an active subscription to something.

Faheem Mitha

Weird, but Ok. But they're going to want to see an example. I guarantee it. Can't you use your current example and overwrite it with gibberish?

Wildcard

@FaheemMitha that might be the bit I was missing.

@FaheemMitha well, I have my current document in two forms. A PDF, which is actually utilized but I don't have a way to change the text of, and an INDD which I can only edit with an active Creative Cloud subscription.

Faheem Mitha

8:36 PM

I mean, they handle some basic top level stuff. But there is extensive scope for customizable. LaTeX is nothing if not customizable.

Read the KOMAscript docs. They're either awe-inspiring or terrifying. Depending on taste.

derobert

@Wildcard I'm trying to imagine what type of checklist this could be where you'd care so much about the layout... Is this something like the checklists pilots use when flying?

Faheem Mitha

@Wildcard You can edit PDFs. At least, with Adobe tools.

Wildcard

@derobert similar in structure. This particular one is for an extensive training lineup.

derobert

You could take the PDF, import a page into gimp as a bitmap, and then just black out any sensitive text

Faheem Mitha

Maybe even with free tools. But they're less reliable.

Or what Anthony said.

@Wildcard I suggest you start with a basic mockup of your thing in LaTeX. Then refine as necessary. These things are usually incremental and iterative, anyway.

Wildcard

8:38 PM

@FaheemMitha I might bite the bullet at some point and make a dummy PDF, just pay for the one month of Creative Cloud to do so.

Faheem Mitha

Did you look at enumitem?

Wildcard

This isn't urgent for me, though.

Faheem Mitha

@Wildcard How much is it? And you can't just buy this InDesign thing?

Wildcard

@FaheemMitha it's the time factor more than the money. It's like $50 a month or something. And no, you can't "buy" it. You can only subscribe.

Faheem Mitha

@Wildcard So this Creative Cloud and InDesign come as a package?

derobert

8:40 PM

@FaheemMitha adobe.com/products/indesign.html

Faheem Mitha

@derobert Oh, it's an Adobe thing?

derobert

yes

Wildcard

@FaheemMitha yep.

Faheem Mitha

So, I'm guessing, expensive to purchase.

Wildcard

@FaheemMitha only with paid tools, by the way. Not with Adobe Reader.

@FaheemMitha impossible to purchase.

They transitioned to only subscription-based software. You can't buy a permanent license for any Adobe software anymore.

Faheem Mitha

8:41 PM

@Wildcard Right. But I thought you might have access to those paid tools.

@Wildcard Ugh.

Wildcard

@FaheemMitha no kidding.

Faheem Mitha

Time to scream and run.

Though my go-to move wrt to proprietary software is to scream and run, anyway.

Wildcard

I do have Creative Suite 6, which was the last version for which you could buy a permanent license. But it only runs on my old Mac.

Faheem Mitha

Though once upon a time I did enjoy working with Sun machines. In the distant past, when the world was young.

TeX is a strange beast. But it will go places nothing else will.

Wildcard

InDesign is a really slick software, I very much like it. But I hate the proprietary lock-in.

Faheem Mitha

8:46 PM

I don't think anyone would call TeX slick. :-)

Needless to say, I've never used InDesign. When I see a word processor, my instinct is to scream and... But I see I've said that already.

I do use LibreOffice occasionally, though. Mostly for reading Word files.

@Wildcard Wrt to TeXnician's comment, you're probably better off using scrartcl, though opinions vary. You also have to like the German approach to document design.

And documentation. Lots of people hate the KOMAScript documentation. I kind of like it. But it might have improved recently.

Wildcard

9:08 PM

@FaheemMitha yeah, InDesign can do a lot lot lot more than just document layout.

Transcript for