In a lot of places, we use read_all() to read the contents of a file into a single String/ByteBuffer. While that may be convenient, it's got issues:

• That means we have to allocate a potentially huge buffer to hold the whole data in memory all the time.
• That means we're unable to process new data as we read it, we just wait until all of the data is read and only then start processing it.
• EOF can never come, and in that case we'll just hang forever.

The alternative is to process new data as it comes, in a streaming fashion. This also means not storing the already processed data in memory.

This mainly concerns parsers and the like. Programs such as copy just have to read all the input to a buffer, to be able to send that buffer to the Clipboard server, so it's fine to use read_all() in those kinds of programs.