I think this misses the purpose of @SEARCH. It's there to perform a path search for executables, like TCC does when you type something at the command line. A function to recursively search arbitrary path(s) for a file should be a different function altogether, say @FFIND.
Quite aware of the current limitations.
I don't use it to search arbitrary paths. But the default is to search the entire current path, which is rather long and represented by a semi-colon list of many directories that do not need to be searched. I just thought that providing a means for the user to search a shorter, user defined "path" would be quicker and more efficient.
My example of searching c:\jpsoft\ and it's sub-directories has become more relevant recently, due to the need to subdivide all the files, which include batch files, into sub-directories to better organize things. Not sure how I see the undocumented function "@ffind" as the go-to function replacing @search. BTW where is @ffind[] help located? Couldn't find it in the off or on line documentation. Probably why I've never used it.
Perhaps a doc page with a few practical examples would be persuasive. I will admit that I have used @search more often than ffind (read almost never) and will do more research to see if I am able to locate what I need with @ffind recursion when I find the docs. I very rarely use plug-ins, just in case this @ffind[] function isn't currently native to tcc.
In any case, the current @search help needs revision as above and adding a better purpose statement. i've been using it the way I think it should be, but I may have missed the point.
Thanks for stimulating some discussion. I was beginning to think everyone was ignoring the suggestions forum. Sometimes there is better information here when users discuss the pros and cons of a suggestion. TTL