>>12606445>brokenone example is the udp library. i worked on a project that almost failed at the last moment because something as braindead basic as a udp library wasn't implemented correctly and was dropping packets like a sieve. matlab support told us udp isn't reliable, but i suspect no one at matworks knew how to fix the issue.
the language itself has several oddities that aren't technically broken, but are extremely questionable design choices. for example the type system is completely fucked, see
>>12600942. This can be a headache because basic commands sometimes can't handle different datatypes, inexplicably. Worse, the weak typing can cause silent errors where data changes shape/structure that can be hard to track down.
the need to pre-allocate memory nullifies a major selling point of the language.
>anti-patternsvectorization (a type of optimization) is critical for performance, and code needs to be structured around vectorizing from the start. premature optimization of programs is a known antipattern.
the script/function distinction has led to over emphasis on scripts, which encourages scripts that become 1000+ lines long (i've seen this from all of my colleagues and then some).
lack of references makes things like linked lists impossible. this makes writing recursive functions extremely difficult, and often odd data structures have to be made to deal with this issue.
etc. etc. these are just ones off the top of my head, not even necessarily the best. i can really tear into matlab's side of i think about all the bullshit i've dealt with over the years.
