HNHacker News
TopNewBestAskShowJobs

heinrichhartman

2,496 karma · joined September 28, 2015

Blog: HeinrichHartmann.com Keybase: https://keybase.io/hartmann Twitter: @HeinrichHartman http://twitter.com/Heinrichhartman GitHub: https://github.com/HeinrichHartmann/
submissionscomments
heinrichhartman··on 0.999...= 1
0.9999 = 1 is a consequence of the way we define rational and real numbers and limits. There are alternative definitions of numbers where this equality does not hold: Non Standard Analysis https://en.wikipedia.org/wiki/Nonstandard_analysis being the most famous one.

But for the sake of argument, let's just define numbers as sequences of digits with a mixed in period somewhere:

    MyNumber := {
      a = (a_1, a_2, ...) -- list of digits a_i = 0 .. 9; a_1 != 0.
      e -- exponent (integer)
      s -- sign (+/- 1)
    }
Each such sequence corresponds to the (classical) real number: s * \sum_i a_i * 10^{i + e}.

We can go on and define addition, subtraction, multiplication and division in the familiar way.

Problems arise only when we try to establish desireable properties, e.g.

(1/3) * 3 = 1

Does NOT hold here, since 0.9999... is a difference sequence than 1.000....

So yes, you can define these number systems, and you will have 0.999... != 1. But working with them will be pretty awkward, since a lot of familiar arithmetic breaks down.

heinrichhartman··on Learning to See in the Dark (2018)
Still -1 second is impressive!
heinrichhartman··on For Donald Knuth, good coding is synonymous with beautiful expression
The hardest part about professional music, is that you have to perform at a schedule. It takes a lot of effort to perpare your body physically and mentally for peak performance ... and often-times you just fail. Have a bad day. That happens.

There will be many concerts, where you are just at 70% of your peak ability.

As a professional musician, you have to make sure your 70%-version is good enough that people gladly pay for seeing it.

heinrichhartman··on Emacs: The most successful malleable system
Yes, I agree that creating more complex extensions is cumbersome. Would love to have ready-to-use python dicts + string functions!
heinrichhartman··on Emacs: The most successful malleable system
> It's not easy by many miles.

Depends on the perspective of-course. I mean, LISP is a real barrier, but once you got over the parenthesis and suck-up some quirks [1], you can get productive quite easily.

Here are two example extensions from my `init.el`.

    (defun hh-curl (url)
      "Create tmp buffer with curl output"
      (interactive "sURL: ")
      (let ((buffer url))
        (with-output-to-temp-buffer buffer
          (shell-command (format "curl -s %s" url) buffer)
          (pop-to-buffer buffer))))

    (defun hh-insert-date ()
      "Insert current date in ISO format"
      (interactive)
      (insert (format-time-string "%Y-%m-%d")))
It's really not all that much syntax. You can get interactive help by navigating to a function call and press `C-h f`. Installation is trivial: Just drop it to init.el.

Use this function my typing `M-x hh <tab>` to see all my extensions, if I forgot the name. I have bound hh-insert-date to a key b/c I use it frequently.

[1] https://www.heinrichhartmann.com/blog/problem-lisp.html

heinrichhartman··on The Webpage, an online RSS reader and news aggregator, styled like a newspaper
Pretty sure it violates copyright of the publishers.

Given that Google is not allowed to show more than the headline and a few words in their search results in the EU, I suspect this is just plain illegal.

This is sad. We should have better laws. Hope you don't get sued.

EDIT: Please let me know if my understanding is incorrect.

heinrichhartman··on Ladder of Algebraic Structures
Categories are Algebraic structures, that are related to this hierarchy:

- Monoids are Categories with a single object.

- Algebras (Non-commutative, Associative) are k-linear Categories, with a single object.

- Any object X in a k-linear category comes with an algebra: R = End(X) = Hom(X,X).

- Any other objects comes with an R-module: Hom(R, X)

- In some cases, we can use this to describe the category as a category of R modules: https://en.wikipedia.org/wiki/Gabriel%E2%80%93Popescu_theore...

heinrichhartman··on Ask HN: Salary negotiations – should I bother getting a competing offer?
Since they have not yet extended an offer to you, it would be foolish to drop out of any other interview processes you are currently involved in.

And even when you have the offer on the table, since you already started the process, I would probably follow through with them in any case.

You are in a much better position if you have competing offers. This is what allows you to negotiate, and leave the company quickly if you don't like it.

heinrichhartman··on Dolt is Git for data
Looks like they are a fork of noms (https://github.com/attic-labs/noms). The object store has the telling name `.dolt/noms`.

Inside are a bunch of binary files. It would be interesting to know more about the on-disk layout of the stored tables.

I was not able to find any documentation. Does someone know more about this? Pointers would be appreciated.

heinrichhartman··on Show HN: Offline text extractor from screenshots in macOS
Very nice!

I found more and more of my workflows getting screenshot driven. It's fast, reliable and accurate. Often-times, I don't care that much about the added overhead of pngs vs text.

E.g. I found the only viable way to export my fb timeline (including pictures and comments), when I left, was to screenshot the entire thing. ~500 images.

Looking forward to having those indexed!

Also really like the software delivery here. No way I would have uploaded that content to a SaaS.

heinrichhartman··on $800M to support small businesses and crisis response
They don't have to. There is no problem. I am glad they are doing something.

However, if you come forward and pledge to provide "800 Million funding and grants" and all you get is Google Ad credits, that's disingenuous.

heinrichhartman··on $800M to support small businesses and crisis response
This is NOT funding:

- $250 million in ad grants ...

- $340 million in Google Ads credits ...

- A pool of $20 million in Google Cloud credits ...

A company that has a Bazillion $ in the bank, has means to really make a difference. This is not.

Edit/Clarification:

The original title was: "$800M funding and grants to support ..". Now they changed it.

I take issue with the claim of providing funding, and all they offer is free ADs. I appreciate that Google is doing _something_. I don't think they have to. I don't think that free ADs is a thing that people need right now.

heinrichhartman··on Getting started with bare-metal assembly
That looks exactly what I was looking for. Thanks!
heinrichhartman··on Lies, Damned Lies, and Averages: Perc50, Perc95 Explained for Programmers
Thanks for your comment!

Relative error is a practical choice, since it allows to cover an extremely large value range (essentially all floating point numbers) with small size ( O(log(range)) ) and zero-configuration. You can't have that with bounding the absolute error.

Also the relative error is what you are interested in most of the time as a practitioner. (200ms+/-10ms; 1year+/-15days)

DDSketch uses relative error for estimation as well.

> If your latency values were clustered around a small range, [...] T-digest will be significantly more accurate.

That is correct!

One point of this example was to demonstrate that merged t-digest can have unbounded errors. In the t-digest paper it was speculated that merged digest have bounded error, but the proof was just more difficult. As it turns out, if you have heavy merges and an extremely large value range, you can get unbounded errors.

heinrichhartman··on Getting started with bare-metal assembly
Does anyone know how to circumvent UEFI?

When the CPU starts, it will start reading instructions from a hard-coded address on the memory bus / EPROM somewhere, right? How can I directly control these bytes?

I don't want some proprietary firmware sit between me and the CPU.

If it's not possible on hardware because "secure boot", or whatever, this should at least be possible in emulators like QEMU.

Does anyone know how to do that? ... or clear up my misconceptions? :)

heinrichhartman··on Münchhausen Trilemma
Just realized you are a student of Marc Nieper-Wißkirchen in Augsburg.

Extend my greeting when you see him! He introduced me to Category Theory around 2006 in Mainz.

heinrichhartman··on Münchhausen Trilemma
It may well be that the search for "truth" has brought us here. However, we found out that "truth" is always a relative. It has to be rooted in assumptions/axioms.

So I would say, it runs on, good choices of axioms and the ability to perform logical deductions.

Set Theory is an attempt to come up with axioms that are so simple that no-one who is not completely out of it's mind would doubt their "truth". This is stuff like:

- There exists a set

- Two sets are equal if all their elements are equal

- We can form subsets from sets, by picking out elements

The only exception is the axiom of choice, which is less intuitively evident, but a technical necessity to build and reason about infinite sets like the real numbers.

heinrichhartman··on Münchhausen Trilemma
Sure, that's the historic route that we ended up at the place we are in.

However, this is not the logical foundations that are used for current teaching and research. There has been a lot of effort, in particular in the late 19the century by folks like Hilbert, Goedel, Frege, etc. to put Mathematics on solid foundations. This is what we ended up with.

heinrichhartman··on Lies, Damned Lies, and Averages: Perc50, Perc95 Explained for Programmers
Shameless plug. Just wrote a paper about this:

https://arxiv.org/abs/2001.06561

Containing a survey of the most popular Latency Aggregation methods used in the industry (Prometheus Histograms, t-digest, HDR-Histogram, DD-sketch/histogram).

heinrichhartman··on Münchhausen Trilemma
The way that Mathematics and approaches this is, using the Axiomatic way.

You can view this as a game:

    Let's make some assumtions / axioms, and see what we can build with it.
The most popular axiom set is Zermelo-Fraenkel-style Set Theory [1], but there are other choices which have been explored.

Note that, there a no claims about absolute truth or reality here. We just play a game in our minds.

The most fundamental question about the game, that we can ask is "Is this game consistent". Goedel found out, somewhat unfortunately, we are not able to answer that question from within the game [2]. However, sofar no intrinsical inconsistencies (paradixies) have been found.

Now, it turns out, that the ZF-Mathematics game is a very useful game, since it allows us to model certain aspects of reality. E.g. Newtons Mechanic, allows us to forecast how planets move in space.

The Natural Sciences (Physics, Chem., Bio) and Engineering are essentially about how to model reality with mathematics. The way to do that is the empirical scientific method [3], where hypothesis are formed, experiments are designed and evidence is collected.

So the question of "truth" does not really arise. It's just:

(A) Mathematics is an interesting mind-game to play.

(B) Mathematics has been used very successfully to model and predict our preceived reality.

That's all you can ask for.

[1] https://en.wikipedia.org/wiki/Zermelo–Fraenkel_set_theory [2] https://en.wikipedia.org/wiki/Gödel%27s_incompleteness_theor... [3] https://en.wikipedia.org/wiki/Empirical_evidence

heinrichhartman··on RFC: Adopt a modern JavaScript framework for use with MediaWiki
At least the content is freely available, so I can stand-up my HTML-only Wikipedia clone if they render the official version unusable with this reactive JS madness.

Turns out HTML over HTTP is GREAT way to deliver encyclopedic content.

heinrichhartman··on Stages of denial in encountering K
Thanks for the clarification.
heinrichhartman··on Stages of denial in encountering K
Looks like the author figured it all out at the very beginning:

> If each punctuation character in the K is a separate part of speech, and you reverse their order

     +      /     !    100
     plus reduce range 100

This seems to be how this works. Compare that to math equations like these ones:

https://en.wikipedia.org/wiki/Maxwell%27s_equations#Formulat...

You would not expect them to read them like a paragraph. You are expect to read them symbol by symbol and unpack from the inner out.

K seems to be a dense language. Expect to read it symbol by symbol, not word by word or line by line.

If will be come more readable if you (1) slow down (2) get used to the programming patterns.

Not saying K is a good language or anything. But the frustration of the author seems to have more to do with impaticence and mismatched expectations than with flaws in the language itself.

heinrichhartman··on Facebook asks for a moat of regulations it already meets
I don't care if "addictiveness" is by design or emergent. Cigarettes were also not designed for cancer.
heinrichhartman··on Building a static serverless website using S3 and CloudFront
I have compile times of >30 seconds with Jekyll blog [1,2]. This is an issue. It adds a lot of friction to the edit/preview loop.

BTW: I just moved it to Netlify since reading this thread. Took me ~5 minutes.

[1] https://github.com/HeinrichHartmann/HeinrichHartmann.github....

[2] https://www.heinrichhartmann.com/

heinrichhartman··on Developer salary spreadsheet compiled from tweets
Nice effort. Was this compiled manually, or by scraping Tweets?
heinrichhartman··on A common mistake involving wildcards and the find command
even better

    printf "<cmd>" | ssh bash
or

    cat cmd.sh  | ssh bash
This passes all of stdin to the remote bash, avoiding any expansions happening on the local side entirely.
heinrichhartman··on A common mistake involving wildcards and the find command
No need for python, this works just as good:

    printf ":%s:" *foo
heinrichhartman··on EGA Chapter 1: The Language of Schemes
Did Grothendieck himself not explicitly object to any re-publishing or translation of his manuscripts?

https://tqft.net/misc/Grothendieck's%20Declaration.pdf

``` Declaration of intent of non-publication

I do not intend to publish or republish any work or text of which I am the author, in any form whatsoever, printed or electronic, whether in full or in excerpts, texts of personal nature, of scientific character, or otherwise, or letters addressed to anybody, and any translation of texts of which I am the author. Any edition or dissemination of such texts which have been made in the past without my consent, or which will be made in the future and as long as I live, is against my will expressly specified here and is unlawful in my eyes. As I learn of these, I will ask the person responsible for such pirated editions, or of any other publication containing without permission texts from my hand (beyond possible citations of a few lines each), to remove from commerce these books; and librarians holding such books to remove these books from those libraries.

If my intentions, clearly expressed here, should go unheeded, then the shame of it falls on those responsible for the illegal editions, and those responsible for the libraries concerned (as soon as they have been informed of my intention).

Written at my home, January 3, 2010,

Alexandre Grothendieck. ```

heinrichhartman··on Ask HN: What Skills to Acquire in 2020?
> The deadlift is certainly dangerous, but only in the sense that almost any large lift is dangerous for the untrained.

Exactly! I, personally, would not do ANY heavy lifts when starting out.

Lot's of great exercises with near-zero risk of injury available. Why start with something that might wreck your spine?

← PreviousPage 7 of 15Next →