Disqus - Latest Comments for k776

Re: Importing vs fetching JSON

Kieran P — Fri, 24 Oct 2025 22:05:33 -0000

I guess one benefit of import over fetch would be for TypeScript types? Imported json could be typed correctly, so you know you won't be calling json keys that don't exist?

This seems like it would be best used to bundle default/fallback translations and fetch other locale only once if it exists, e.g.

import english from './en.json' with { type: 'json' }
const locale = navigator.language.slice(0, 2)
const { default: other } = await import(`./${locale}.json`, { with: { type: 'json' } });
console.log(other?.hello || english.hello)

Re: Update millions of records in Rails

Kieran P — Thu, 27 Apr 2023 05:55:20 -0000

Re: #2, or even better, bypass Rails completely. SQL is perfectly capable of doing this itself.

UPDATE addresses SET city = LOWER(city), state = LOWER(state);

Should be many times faster than doing things the Rails way.

Re: Speeding up views rendering in Rails 4

Kieran P — Thu, 23 Jun 2022 21:11:36 -0000

You could possibly speed it up further by caching often used views in memory, rather than doing a File.binread each time it needs to be rendered.

Re: Small nice feature that emerged in Ruby 3.1... But has a nasty quirk

Kieran P — Sat, 11 Dec 2021 01:05:23 -0000

What about using hash brackets instead?

p {a:,b:}

In that case, it ruby smart enough to determine is as a hash, or does it parse it as a block?

Re: OPAQUE: The Best Passwords Never Leave your Device

Kieran P — Wed, 09 Dec 2020 02:32:36 -0000

What about a simpler approach? Client JS on login/signup page does `SHA256(password)` and sends the result as the password to the server, which adds an additional encryption on top (bcrypt) before storing in the DB?

That way the real password is never plain text over network requests, server never sees original password, only the hash, but server still has a way to validate the resulting hashed password. And it doesn't require an additional network request.

Going a step further, browsers could implement something for this in the input tag. e.g. `<input type="password" name="password" hashed="SHA256"/>`, and when a form is submitted, the password is hashed before making the network request without needing any JS. Then browsers could give security warning when a page contains a password field that won't be hashed.

Re: Ruby vs Crystal Performance – Pavel Timofeev – On software development, startups, and other things

Kieran P — Fri, 12 Jun 2020 03:05:18 -0000

You can also add the --release flag when building crystal for additional optimisation.

Re: Cloudflare for SSH, RDP and Minecraft

Kieran P — Mon, 13 Apr 2020 18:23:56 -0000

Fantastic. The only draw back is that it only allows one SSH IP per account. In our case, we have several bastion boxes for different parts of the application under the same domain, so we can't protect them all using Spectrum.

Re: SendGrid DNS White Labeling and CloudFront for Secure Click Tracking Links

Kieran P — Mon, 11 Sep 2017 22:17:09 -0000

Hey Matt. In your diagram you refer to "yet_another_hash.sendgrid.net" <- Where do you get this value from? I have setup a Cloudfront distribution, setup the origin and behaviour, and contacted SendGrid, but it is not operating properly. Just wondering if it's because I couldn't find the "yet_another_hash.sendgrid.net" value you referred to and had to use just "sendgrid.net". Any thoughts?

Re: Updated NGINX Plug-In for New Relic

Kieran P — Sun, 08 May 2016 18:50:44 -0000

Hey Ted. Did you ever find a fix? We've just run into this issue with one of our hosts. The daemon keeps crashing when it encounters whatever is causing this, which leaves us blind as to how many connections that server is handling.

Re: Generating Complex PDF Documents in Rails with Prawn

Kieran P — Sat, 21 Mar 2015 15:45:55 -0000

Everything you mentioned is possible with wicked_pdf... See https://github.com/mileszs/... - Under the advanced options sections, you can set :margin, :header and :footer settings which take effect on all pages, working exactly as it should. You can also use some page-break css classes to split pages where you want them split. A lot of control there...

Re: Generating Complex PDF Documents in Rails with Prawn

Kieran P — Sat, 21 Mar 2015 02:02:11 -0000

I've found it much easier to use a tool like wicked_pdf, which uses wkhtmltopdf to generate the PDF files. You can design your PDFs with HTML and CSS in ERB files. So much easier. Not sure I'd switch away from that to something like Prawn, which seems far more complex.

Re: Choosing the right data store

Kieran P — Wed, 30 Jul 2014 01:43:06 -0000

Where is PostgreSQL? On http://db-engines.com/en/ra..., Postgresql is at #4, while MariaDB is at #28 and Percona is at #80. Why have you not added support for Postgresql early on?

Re: The New Rubygems Index Format

Kieran P — Mon, 24 Mar 2014 21:38:31 -0000

Very interesting. Why not use versions for the names autocomplete though (other than the size aspect) and have one less file to append to, and the benefit of autocompleting version as well?

And how would bundler go about request multiple dep files at once, or does it need to hit /deps/GEM_NAME for each gem until all deps are met? Could the /deps/ endpoint accept multiple gem names and transmit them in a single file for Bundler by appending all single deps together separated by a "== GEM_NAME" header perhaps?

I'd been interested in knowing the total size of the old index v.s. new index too. I didn't notice any size comparisons.

Re: Revocable sessions with Devise

Kieran P — Sun, 16 Feb 2014 23:03:02 -0000

Here is a modified version which provides an upgrade path (all users aren't logged out when the code is deployed), and also updates an accessed_at timestamp on the users record. It also has a little less code in the user model. https://gist.github.com/Kie...

Re: SSL Certificate Types

Kieran P — Sat, 16 Nov 2013 04:37:57 -0000

Note: You cannot get organisation or extended validation SSL certs with wildcard. Those can only be single name, which is a real pain :-( They are also far more expensive.

Re: Realtime Strategy Game 0 A.D. Enters Final Crowdfunding Stretch

Kieran P — Tue, 01 Oct 2013 18:58:42 -0000

We aren't paid developers. Most of us are part time volunteers. We currently have one paid contractor who is working on performance and his contract ends soon. The fundraiser will be used to extend his contract period.

Re: Realtime Strategy Game 0 A.D. Enters Final Crowdfunding Stretch

Kieran P — Tue, 01 Oct 2013 18:57:05 -0000

You're paying for a better end result, something more stable, better performing, better looking...

Re: Realtime Strategy Game 0 A.D. Enters Final Crowdfunding Stretch

Kieran P — Tue, 01 Oct 2013 18:53:41 -0000

True, tax deductions are not applicable outside the US, but then Kickstarter doesn't even have the option for it at all, so some tax deductibility with IndieGoGo for those inside the US is better than nothing.

"Another US only campaign" <- That isn't at all accurate. It's a US donation perk, but nothing stops people from other countries from donating, it just isn't tax deductible for them...

Re: 10 Years In Alpha: How Volunteer Modders Created 0 A.D. With The Help Of A Community

Kieran P — Tue, 01 Oct 2013 15:51:54 -0000

To be fair though, the game was rewritten from scratch in 2009, so 10 years in the making perhaps, but the code itself is only 4 years old.

Re: Realtime Strategy Game 0 A.D. Enters Final Crowdfunding Stretch

Kieran P — Mon, 30 Sep 2013 19:45:55 -0000

The reason we went with IndieGoGo over Kickstarter is that Indie supports non-profit organisations, which the company behind 0 A.D. is. This means that not only do we get lower fees on the campaign, but we don't need to worry about tax related issues... all donations are going to the non profit organisations account (SPI), and we can offer our US donators tax deductible donations.

Re: Free RTS Game 0 A.D. Sees New Alpha Release ~ Web Upd8: Ubuntu / Linux blog

Kieran P — Fri, 06 Sep 2013 22:55:25 -0000

No. All the developers contribute their free time. A campaign is running now to raise funds to hire first a developer, then an artist, to help complete these game within a year.

Re: The Curious Case of ActiveRecord Matching Yesterday's Events Even Though I Told it Not To

Kieran P — Tue, 19 Feb 2013 15:37:21 -0000

Or the simplest solution: Never use a date field. Always use a datetime field, and then when you pull it out of the database, use `start_time.to_date`.

Then you just do `Event.where("start_time > ?", Time.zone.now.beginning_of_day)`

Re: Phusion Passenger 4.0 beta 2: Syscall failure simulation framework, focus on stability

Kieran P — Sun, 10 Feb 2013 19:02:42 -0000

Opps. Looks like something went wrong. Response times skyrocketed after upgrading from beta 1 to beta 2. See this ticket: http://code.google.com/p/ph...

Re: Hobo 2.0.0.pre7

Kieran P — Fri, 07 Dec 2012 05:22:53 -0000

Please make hobo_fields Rails 3.3 compatible. It's holding me back from upgrading.

Re: Making your ruby fly

Kieran P — Sat, 24 Nov 2012 14:12:31 -0000

This is excellent. Things are so much faster now. Using "rvm_configure_env=(CFLAGS="-march=nocona -O3 -fomit-frame-pointer -pipe")" in ~/.rvmrc, the ruby iterator test case took 7.5s before, and only 3.2s after. Other things show similar speedup. 'rails runner' from 12.4s down to 6.6s, 'rake routes' from 12.5s down to 6.8s, and 'rspec spec' from 5m 53s down to 3m 17s. So an amazing savings.