Not quite a bug but a confusing behavior: Sometimes throws vague error if two blocks of code are not separated by a semicolon

[Spederan]

I was confused because my files ran fine both minified previously and they were currently running fine nonminified, but when i converted each to a string and added them together separated by a space (which worked fine dozens of times before across dozens of files), it threw an unexpected and vague error. After 20 minutes, i finally figured out its because it thought i was missing a semicolon, in retrospect it must have interpreted the code as a single line of code, and the missing semicolon as an error. My solution is now separating each file with a semicolon and a space, just to make sure.

At first i thought it was a bug because my original files ran fine, but my minified files (the minification process i thought i had perfected before) did not run fine. Now i am unsure, because after thinking about it this might have to do with javascript itself being confused, after i put the code into a single string which it would interpret as a single line.

My issue in relation to UglifyJS is that the error message didnt really point to anything but UglifyJS's own internals. I had no idea what was going on. An error message that points out the error in the string it was parsing wouldve been massively helpful, and i would have immediately seen the problem. For example: func() let a = 5; would throw an error i believe, and a more helpful error message would include that segment in the message. Also if there was a way to detect it was specifically missing a semicolon, that would have been helpful also. And in theory i dont see why a semicolon cant be automatically inserted here, as clearly the only context in which it makes sense is one in which it exists.

For reference, here is the error message i got:

JS_Parse_Error [SyntaxError]: Unexpected token: name «let», expected: punc «;»
at JS_Parse_Error.get (eval at (/home/anonymous/Documents/Spederan/speedrunjs/node_modules/uglify-js/tools/node.js:18:1), :63:23)
at getStackString (node:internal/util/inspect:1241:37)
at formatError (node:internal/util/inspect:1363:15)
at formatRaw (node:internal/util/inspect:986:14)
at formatValue (node:internal/util/inspect:841:10)
at inspect (node:internal/util/inspect:365:10)
at formatWithOptionsInternal (node:internal/util/inspect:2273:40)
at formatWithOptions (node:internal/util/inspect:2135:10)
at console.value (node:internal/console/constructor:340:14)
at console.log (node:internal/console/constructor:377:61) {
filename: '0',
line: 6610,
col: 17,
pos: 365301
}

Not sure if this counts as a bug or a feature request, but i wanted to share it regardless.

[alexlamsl]

Short answer: use the CLI for what you want:

$ echo "f() g()" | uglifyjs
Parse error at STDIN:1,4
f() g()
    ^
ERROR: Unexpected token: name «g», expected: punc «;»
    at JS_Parse_Error.get (...

Basically there isn't a standardised way in JavaScript to display SyntaxError, so almost all third-party tools would handle custom errors like the ones thrown by uglify-js poorly. They won't even print a stacktrace without additional hacks that has been put in place (though the one you've shown above seems odd, investigating...)