HNHacker News
TopNewBestAskShowJobs

alexhutcheson

3,780 karma · joined February 17, 2013

Nothing I post reflects the views or policies of my employers, past or present.
submissionscomments
alexhutcheson··on SDL Moves to GitHub
> It is a symptom of the fact that the FSF and other FOSS organizations have displayed a fantastic failure of technical leadership for over 3 decades.

I agree that many FSF/GNU projects have made horrible technical choices, but I don’t think that’s really the root cause here. The reality is that good software requires an enormous amount of skilled labor to create and maintain, and the FSF and other OSS orgs can only muster a small fraction of the engineering hours that a mid-sized for-profit software company could.

In addition, good user-facing software requires more than just good engineering - you need good UX, product direction, etc. Sometimes the creator or maintainer of an OSS program will have the good fortune to be reasonably good at all of these, but it’s very rare, and there aren’t nearly as many UX designers or product managers looking to contribute to open source as there are developers. Even if someone did want to help in this role, the developers on most projects would probably ignore them.

alexhutcheson··on Don't Offer a Free Plan
If you use it as a full project management tool it can be a bit painful, but is there a better tool you’d recommend for someone who just wants a good bug tracker?
alexhutcheson··on The Dunant subsea cable
http://dls.virginia.gov/commission/materials/subseacables.pd...
alexhutcheson··on The effect of switching to TCMalloc on RocksDB memory use
Unless those projects have required copyright assignments (neither does, to my knowledge), then you’d need to get agreement from the copyright holder for every contribution that still has code in the project. If there are any holdouts (or even any people you just couldn’t get in touch with) then you’d need to remove the parts they contributed to proceed.

More practically, anyone who was asked could reasonably respond “Why do you need to own the copyright? It’s already permissively licensed (in jemalloc’s case, with an LGPLv2.1-compatible license) - just use it!” FSF policy is the only obstacle, and they could make an exception if they wanted to.

alexhutcheson··on Quantitative Economics with Julia
If you’re interested in quantitative economics, I would highly recommend the “Metrics Mondays” series of blog posts by Prof. Marc Bellemare:

http://marcfbellemare.com/wordpress/metrics-mondays

They have a lot of pragmatic rules-of-thumb and best practices for doing causal inference on real data.

alexhutcheson··on The effect of switching to TCMalloc on RocksDB memory use
The FSF requires copyright assignment for code added to glibc, which makes it difficult to add code from other projects.
alexhutcheson··on The effect of switching to TCMalloc on RocksDB memory use
The FSF requires copyright assignment for code added to glibc, which creates an impediment to including code from other projects.
alexhutcheson··on Launch HN: Opstrace (YC S19) – open-source Datadog
One pain point with Prometheus is that is has relatively weak support for quantiles, histograms, and sets[1]:

- Histograms require manually specifying the distribution of your data, which is time-consuming, lossy, and can introduce significant error bands around your quantile estimates.

- Quantiles calculated via the Prometheus "summary" feature are specific to a given host, and not aggregatable, which is almost never what you want (you normally want to see e.g. the 95th percentile value of request latency for all servers of a given type, or all servers within a region). Quantiles can be calculated from histograms instead, but that requires a well-specified histogram and can be expensive at query time.

- As far as I know, Prometheus doesn't have any explicit support for unique sets. You can compute this at query time, but persisting and then querying high-cardinality data in this way is expensive.

Understanding the distribution of your data (rather than just averages) is arguably the most important feature you want from a monitoring dashboard, so the weak support for quantiles is very limiting.

Veneur[2] addresses these use-cases for applications that use DogStatsD[3] by using clever data structures for approximate histograms[4] and approximate sets[5], but I believe its integration with Prometheus is limited and currently only one-way - there is a CLI app to poll Prometheus metrics and push them into Veneur[6], but there's no output sink for Veneur to write to Prometheus (or expose metrics for a Prometheus instance to poll), and you aren't able to use the approximate histogram or approximate set datatypes if you go that route, because they can't be expressed as Prometheus metrics.

It would be extremely useful to have something similar for Prometheus, either by integrating with Veneur or implementing those data structures as an extension to Prometheus.

[1] https://prometheus.io/docs/practices/histograms/

[2] https://github.com/stripe/veneur

[3] https://docs.datadoghq.com/developers/dogstatsd/

[4] https://github.com/stripe/veneur#approximate-histograms

[5] https://github.com/stripe/veneur#approximate-sets

[6] https://github.com/stripe/veneur/tree/master/cmd/veneur-prom...

alexhutcheson··on What’s Wrong With The Honeywell Humidifier
Which ones?
alexhutcheson··on New Xquartz release with native Apple Silicon support
Wow. The previous release was in 2016, so I was sure that XQuartz was dead for good. Nice to be pleasantly surprised!

I occasionally use XQuartz to run graphical programs over SSH using X forwarding, although I’ve mostly moved on to other approaches due to a combination of:

1. Bad support for HiDPI displays (theoretically fixable in XQuartz - curious if they will tackle it now).

2. Annoyingly high input and redisplay latency (probably not fixable, just an inherent property of the very chatty X protocol).

Still nice to have the option.

alexhutcheson··on Johnson and Johnson single-shot vaccine appears 66% effective in global trial
That’s just semantics around the word “scheduled”. The FDA insisted on a course of action that makes it unlikely the AZ vaccine will receive an EUA before April. They did this despite the fact that pretty much no one with vaccine expertise doubts that it will meet the safety standards or the (relatively low) efficacy standards required to grant that EUA.

No one disputes that AZ screwed up their trial, but it doesn’t make sense to prevent people from accessing a safe and reasonably effective vaccine just to make a point about that. That’s why the UK and EU have approved it and are moving forward with distribution, while the US waits months for absolutely perfect data before doing anything.

alexhutcheson··on What's New in PHP 8.1
> Even CPython is faster.

The benchmarks I’m familiar with[1] show recent PHP releases to be a lot faster than CPython for many tasks. Are there better benchmarks I should be aware of?

[1] https://benchmarksgame-team.pages.debian.net/benchmarksgame/...

alexhutcheson··on Johnson and Johnson single-shot vaccine appears 66% effective in global trial
We're limited by supply, so the existence of (limited quantities of) better vaccines isn't a good reason to block people from choosing to get a safe vaccine that still provides a reasonable degree of protection.

There are plenty of people in the lower-risk categories who would prefer to get a >60% effective vaccine now, rather than wait months for their turn in line for vaccines with the highest efficacy.

alexhutcheson··on Johnson and Johnson single-shot vaccine appears 66% effective in global trial
The AstraZeneca vaccine has been approved for use in the UK since December, but the FDA isn't currently scheduled to approve it in the US until April.
alexhutcheson··on The Evolution of the QEMU Translator
Rob Landley was at one point working on an interesting project to combine a fork of the Tiny C Compiler (tcc) with QEMU's Tiny Code Generator (TCG) to create a compact, BSD-licensed drop-in replacement for gcc and binutils that would work on all the platforms supported by TCG:

https://landley.net/qcc/

http://landley.net/hg/qcc/file/tip/todo/todo.txt

It seems like a pretty cool idea, so I'm hopeful that eventually he'll have the chance to hack on it and get it working.

alexhutcheson··on Looking for More Debugger
> Trying to have some debugger program like gdb and then talking to it from emacs or a terminal or something will be a big pain.

If you're already familiar with Emacs commands, then the M-x gdb interface is as good as any debugger interface in an IDE.

If you're not already an Emacs user, then the TUI[1] (gdb -tui) is surprisingly good - much better than the "old school" approach I was taught in college of just interacting via the gdb prompt.

> Debuggers work well for what, windows and websites?

I spend most of my time on a large C++ project that does neither of these things, and I still use gdb a lot. In my experience, it's most useful for unit tests - you identify a bug, write a unit test that triggers the bug, and run the unit test under gdb to poke around and understand what's going on.

gdb and lldb don't work great when run on the whole program, mainly because it's hard to specify the actual behavior you want when debugging a system where multiple threads are executing concurrently. Almost all of our unit tests are single-threaded, so this isn't a problem when debugging unit tests.

If you're working on a C or C++ program and using a supported processor, I also highly recommend trying out rr[2], the record and replay debugger, which you mentioned in your comment. Unlike gdb, it's reasonably easy to use and understand even if your program has lots of concurrency, and it pretty much eliminates the need to do "printf" debugging, since you can print variables from any line in your program without recompiling and re-running. You can still use the Emacs or TUI gdb interfaces when using "rr replay".

[1] https://sourceware.org/gdb/current/onlinedocs/gdb/TUI.html#T...

[2] https://rr-project.org/

alexhutcheson··on AMD Ryzen 9 5980HS Cezanne Review: Ryzen 5000 Mobile Tested
> There are reasons why Nvidia don’t try to compete on this market.

Maybe not at the high-performance end, but the Nintendo Switch is powered by an Nvidia Tegra X1, and they won that contract largely by proving the concept in their own Nvidia Shield “microconsole”.

alexhutcheson··on Our Journey to PostgreSQL 12
It’s a dating app.
alexhutcheson··on Raspberry Pi Enters Microcontroller Game with $4 Pico
Link? The dev boards I’m familiar with are $10. Not that I really care for hobby purposes - I’m not buying 100s of them.
alexhutcheson··on Android Ported to RISC-V
I think you’re likely underestimating how complex a modern microarchitecture like the A78 or X1 is compared to the RISC-V cores that have been open sourced so far. It’s unclear to me what organizations would have the resources to develop something like the A78 for RISC-V and would also be happy giving it away for free.

Note that in the ARM world everyone except Apple and Nvidia have given up on developing their own microarchitectures - they all just license one from ARM and maybe apply some customizations on top.

alexhutcheson··on Android Ported to RISC-V
That might be true, but I wouldn't expect the microarchitectures that achieve this performance to be free (in either the "gratis" or "libre" senses of that word). Designing a superscalar application processor requires an enormous amount of work by engineers who are experts in their field, and that work needs to be compensated somehow.

The ISA may be open, but there is no requirement for the microarchitecture implementing the ISA to be, and it's hard to imagine anyone doing the work to produce a performance-competitive chip and then giving it away. Even if there is a "consortium" model where multiple companies benefit from a single design, they would likely set up some sort of licensing arrangement to prevent free-riding by companies that did not contribute to the design.

alexhutcheson··on Cgit, Nginx and Gitolite: A Personal Git Server
If you just want to browse some repos via a read-only web interface, you might also want to check out gitweb[1] and the wrapper script git-instaweb[2], which are bundled with git.

[1] https://git-scm.com/docs/gitweb

[2] https://git-scm.com/docs/git-instaweb

alexhutcheson··on Cgit, Nginx and Gitolite: A Personal Git Server
Another option is to run a web GUI like cgit, gitweb, Gitea, or Gogs without exposing it to the internet (so it's only accessible at http://localhost:8888 or similar).

When you want to view it from another machine you can just use SSH port forwarding, like:

    ssh -L 8888:localhost:8888 user@hostname
alexhutcheson··on Raspberry Pi Enters Microcontroller Game with $4 Pico
That's just the chip, not the board.
alexhutcheson··on Raspberry Pi Pico and RP2040 Microcontroller
If they're sticking with ARM I would bet on them using an ARM Mali GPU, since they would almost certainly be licensing the microarchitecture from ARM anyway.
alexhutcheson··on Raspberry Pi Pico and RP2040 Microcontroller
It has two cores, so I don’t believe disabling interrupts would be sufficient unless you’re only using one core.
alexhutcheson··on Red Hat announces no-cost RHEL for small production environments
This page has some interesting details: https://linux.oracle.com/switch/centos/
alexhutcheson··on Linux Running on Apple M1
In that article, Torvalds specifically says:

"The main problem with the M1 for me is the GPU and other devices around it"

That's a separate issue than the ISA extensions, which apply to the ARM CPU.

alexhutcheson··on Red Hat announces no-cost RHEL for small production environments
Fermilab discontinued work on Scientific Linux in 2019 in favor of using CentOS: https://listserv.fnal.gov/scripts/wa.exe?A2=SCIENTIFIC-LINUX...

It’s possible they could choose to restart work on Scientific Linux, but nothing has been announced yet.

alexhutcheson··on Red Hat announces no-cost RHEL for small production environments
Those are probably the users that RedHat is trying to nudge onto RHEL. "Enterprise" software is normally quite expensive, so the marginal cost of some RHEL Server licenses is probably a small fraction of the total cost of deploying the software in most cases.

More generally, orgs that are running "Enterprise" software are already set up to pay for software. Going from spending $X to spending $X+Y is often not a big deal ("Ok, I'll run that through the same process I used to get approval for buying FooSoftWorks"), while going from $0 to $Y can be very hard ("I don't even know who I'd need to ask to get purchase approval for that").

← PreviousPage 4 of 29Next →