Struct SearchCtx 

pub struct SearchCtx<'a> {
    pub matcher: &'a RegexMatcher,
    pub sep: char,
    pub globs: Option<&'a Globs>,
    pub max_count: Option<u64>,
    pub drop_cache: bool,
    pub count_mode: bool,
}

Everything a search needs that is the same for every archive in a run.

Borrowed and shared across the whole scheduler: it is Sync, so one instance is read by every worker. The per-job values — the path, the Searcher, the printer, and the inflate strategy — stay as arguments, because each worker must own its own (Searcher is not Sync) and the strategy is chosen per file from its size.

Fields

matcher: &'a RegexMatcher

sep: char

globs: Option<&'a Globs>

max_count: Option<u64>

drop_cache: bool

A request, not a decision: open_decoded_dropping narrows it to archives, so a plain live logfile is never evicted.

count_mode: bool

Whether the run is in -c mode. Not used to choose a printer — cli already did that — only to decide whether a suppressed count is worth a note on stderr.