Half, but not exactly what you want either, can be done with PICK from count. My guess is that anything else needs some code wrapped around a sipping pipeline.
I think you might have more luck trying to reduce the xcompare output down to something manageable. [The pipeline regression test generates close to 100,000 lines of output these days. SuperC still copes nicely.] On 31 August 2011 09:18, DUGALEIX Michaël <[email protected]> wrote: > Hello, > > I want to compare two rather similar too-large-for-a-human files (20000 > lines). > These files are the traces left by a trace('i')'ed REXX before and after a > (simple) modification. I'm looking for regressions in the program. > > A COMPARE or XCOMPARE works perfectly (is never lost), but gives me too many > differences. > So, I'd like to filter the traces a bit to have less lines to read/compare. > > I can easily isolate the few lines I modified in the REXX (locating a given > string), but I'd also like to see the few lines following my given string > (to look for regressions). > > Is there a filter to do that (something like "locate /theString/ take 3 > after"), or a combination of filters (locate, take, combine, strfrlab, > ?...) ? > > For example, with this input : > > A > B > C > DtheString > E > F > G > H > I > J > KtheString > L > M > N > O > P > Q > R > S > U > V > W > > I want this ouput : > > DtheString > E > F > G > KtheString > L > M > N > > Thanks > Michaël >
