[Theo Verelst]

I remember from the earlier days of tcl, when I was looking around for at least a decent interpreted language, preferably with user interface component for a HP UX (hp720) based job, for which I had made some X stuff but with quite old widgets and without much of menus, that the tcl interpreter used 4 bytes (an integer) for storing a character.

Come to write about it, I don�t know wether I gathered that would be for storing just one char in source or var, I guess I somehow assumed that was the general idea.

Anyone who knows wether that is the case at present ?

[NEM] - Do you mean storing a char as in
 set x a  ;# Set x to the character 'a'
?

In this case, Tcl stores a string of length 1. So, you now have the size of the Tcl_Obj structure (24 bytes?) plus the size of the string representation of that character in UTF-8, so quite a bit more than 4 bytes used now.

[TV] Sort of what I meant, but I remember something about string being not stored very efficiently for instance in a text widget for reasons of tags and labels, which is often fine, of course.

And also when we store a list, how much overhead is there on top of the actual characters which make up for an uncompiled or uncompressed list was the line of thinking.

[Lars H]: The [Compact data storage] page was set up for discussions of how to compactly store data. The related [Measuring memory usage] page contains some scripts that one can use to measure how much memory is being used for storing various kinds of data.