Using markdown in post reply crashes Firefox

[ivanistheone]

When trying to post a reply to this thread, I typed in the following text:

Good post. I was able follow most things despite being a little scared of DI due to past experience with Java Spring. You promised "There are no complex libraries to install. There is no black-box framework running your code." which make me interested.

Here is some feedback

1. Third paragraph in the Intro, I had trouble interpreting "Despite Al/ML hype and adoption skyrocketing, ..." since I wondered if you mean ML programming (like building software that happens to include ML), or using AI/ML (i.e. LLMs) to write software (e.g. business software). I'm not sure if this needs to be fixed, but with "vibe coding" and ChatGPT taking over in the news, you might want to rephrase to clarify you're talking about humans doing traditional coding where in the domain of Al/ML.

The first section "Example: Trainer class" was great, easy to follow + keeps the promise "There is no black-box framework running your code." from the Intro.

2. The second section "Building configurable objects" lost me a bit.
- The code example is very long (could you cut down some parts at the cost of losing realism)
- The use of "dacite" in the end is confusing for someone who doesn't know about the library (me). Would there be a way to stick to plain Python, w/o bringing in dacite? (to keep the promise from the Intro)
- If you choose to keep 'dacite', you should also show the contents of 'config-yam!' so we get the full picture

3. The third section of Anti-Examples was pretty good, although I would suggest the following changes:
- Add a paragraph explaining what readers will see in the "Anti-Examples" section, e.g. "Sometimes it's best to learn by example, so I'll now show examples of ..."
- Each subsection needs a little intro (explain in words what you're about to see)
- In each subsection, put the code first, and the bullet points explanations after the code (otherwise the points don't make sense, since I haven't seen the code yet). You did this already in the "Composition over inheritance" example, so change the others to be the same.

4. Last but not least, I think "..an implicit dependency which each subclass is forced to inherit, if it won't use it." should be "an implicit dependency *that* each subclass is forced to inherit, *even* if it won't use it."

If you would like more feedback on this post (or on future posts), consider posting them in the writing review' channel on Zulip: https://recurse.zulipchat.com/#narrow/channel/21709-writing-review/ There are usually lively discussions around blog posts and/or drafts of such.

then I wanted to see if the bold worked correctly so I pressed Preview button and the browser tab froze.... like hard!

Firefox 136.0.2 (aarch64) on macOS

I just tried again, and clicking POST also hard freezes the tab.

I'm going to try to reproduce this to see exactly which subset of the markdown is causing this, but posting the full text first.

[ivanistheone]

Okay this is the longest chunk that doesn't crash the tab:

Good post. I was able follow most things despite being a little scared of DI due to past experience with Java Spring. You promised "There are no complex libraries to install. 

1. Third paragraph in the Intro, I had trouble interpreting "Despite Al/ML hype and adoption skyrocketing, ..." since I wondered if you mean ML programming (like building software that happens to include ML), or using AI/ML (i.e. LLMs) to write software (e.g. business software). I'm not sure if this needs to be fixed, but with "vibe coding" and ChatGPT taking over in the news, you might want to rephrase to clarify you're talking about humans doing traditional coding where in the domain of Al/ML.

The first section "Example: Trainer class" was great, easy to follow + keeps the promise "There is no black-box framework running your code." from the Intro.

2. The second section "Building configurable objects" lost me a bit.
- The code example is very long (could you cut down some parts at the cost of losing realism)

it is exactly 1000 characters

adding even a single word to it produces the crash

So there is something wrong happening around the 1000 char limit.

[ivanistheone]

Update: I was able to post the whole reply w/o problems using Chrome, so this is a Firefox-specific issue.

[ivanistheone]

Update 2: that thread won't even load anymore in Firefox.

A spinner shows up and the browser tab is unresponsive:

I tried turning off all plugins, but the problem persists.

[nicholasbs]

Thanks for the report and troubleshooting!

I was able to reproduce on Firefox, but not Chrome or Safari, so it's definitely a Firefox-specific issue.

I'm not sure when we'll have time to investigating/debug further; I assume it's an issue with the Markdown rendering, but that's just a guess. If anyone has time and interest, all the relevant code here should be open source.