When FLAC means literally "free lossless audio codec", mustn't we assume that there is even no specification, no method for lossy compression..? According to this, there should be no way to use a format that by definition preserves a bitstream to produce a lossy output that cannot be restored to the original bitstream. If the produced FLAC cannot be restored to the complete source, then the software must be at fault. It's as if a zip or rar file couldn't produce the original file, it would equally result in a CRC error, only that the usual FLAC encoder will likely skip the such a check.
But the reason why I'm bumping this thread is because I started today to take a closer look at my FLAC files, for fear of getting fake FLACs. We all know by now the widespread expectation, sometimes even pressure, to provide FLAC files, and I'm sure within a sharing community such as Soulseek and others, having FLAC versions of something can mean the difference between getting access to another person's shares or being denied access. Some people may be tempted to convert their MP3s en masse to FLAC, and some albums seem indeed to be converted MP3s. Herein lies the danger: having invested time and effort into updating your collection from MP3, only to later learn that you had been fooled. Now you may not even have the original MP3 files anymore, and you're left with fake FLACs that take up precious disk space unnecessarily, and you're facing having to invest even more time into either getting back proper MP3s, or true FLACs. Also, there is always the danger of discarding a high-quality MP3 (good, clean vinyl with proper recording technique @ 320 kbps) in favor of a flawed FLAC (scratched vinyl etc.). I'll be keeping my MP3 versions unless I can be reasonably sure about the quality of the FLAC.
I've downloaded some tools to check the audio properties of files, such as
Lossless Audio Checker
https://losslessaudiochecker.com/which scans your FLAC for signs of upscaling, upsampling and transcoding.
and Spek
http://spek.cc/showing you a spectogram of an audio file (frequency range), revealing, among other things, telltale cutoff points of lossy encodings.