tiny-http VS node-ipc

kcgi describes itself as being "minimal" whereas actix-web calls itself "powerful", "extremely fast" and as having "a lot of features out of box"... seems about as fair as comparing the 320MB Apache HTTP Server repo against the 52KB tiny_http library

I'm developing the backend of https://www.cozydate.com/ in Rust. Async Rust is not productive yet, so I tried rouille http server which lets me write non-async request handlers. Unfortunately, it uses an unbounded thread pool and falls down under load https://github.com/tiny-http/tiny-http/issues/221 . Then I tried Tide and a threadpool to call my non-async API handlers. This worked, but was really ugly, and I had issues with uploads after deploying to Heroku https://github.com/http-rs/tide/issues/878 .

tiny-http is small. You'll probably be able to make something with less than 1 Mb.

That's not a simple http server, something like tiny_http would be with only... 17 total dependencies by default. Actix is a full framework with an abundance of features, and correspondingly-large dependency tree.

It should be noted that when developers have the ability to upload packages themselves, there is no guarantee that what the user gets corresponds to the source code that is available. The malware in node-ipc is suspiciously absent from the published source repository which has the latest release in 2021. If you were pulling directly from the source code you would not have seen this. This is an issue with npm, not libre software or "open source."

Direct link to the github discussion: https://github.com/RIAEvangelist/node-ipc/issues/233 (many comments have been removed)

Actually, to my knowledge, I was the very first person to discover the malware contained in the commit range of 847047cf7f81ab08352038b2204f0e7633449580 -> 6e344066a0464814a27fbd7ca8422f473956a803

From the article:

> I don’t think this can be classed as open source anymore:

> The definition of an Open Source License is quite clear:

> 5. No Discrimination Against Persons or Groups The license must not discriminate against any person or group of persons.

> I don’t really want to have to read through each of my dependencies and transitive dependencies licences to determine whether I am agreeing to discriminatory terms by using a library.

I think the author of the article has misunderstood the definition, thus reached the wrong conclusion.

The non-discrimination rule applies only to accessibility and nothing else. Simply put, you provide the same code/product to everybody, including Satan Claus and Mr Putin under the same set of condition and permissions. Adding/removing malicious code does not change the fact that the code is by definition open sourced.

All and all, this is not a license problem.

Now, talk about node-ipc, which just got attacked by mobs (see https://github.com/RIAEvangelist/node-ipc/issues?q=is%3Aissu...).

If an open source project is a scam, then it's a scamware. If an open source project is malicious, then it's a malware.

Personally, as a normal human being, it is hard to keep a peace of mind after watching how the Russians fired multiple heavy rounds to kill the elderly couple who just traveling peacefully in a car down the road near a hospital. It is even harder to keep a peace of mind after watching a video recorded by a son showing how the Russians shoot and killed his father who sits in the driver's seat right beside him. I fully understand and respect the anger.

However, I do agree that people need to be more mature on this even during this difficult time. Turn your project into a malware only hurts your own reputation and people who trusted you. Once the trust is gone, it might never recover. There are many ways to actually hurt those who contributed the invasion. Be constructive and accurate, or at very least don't be destructive.