levisonm@qucis.queensu.CA (Mark Levison) (03/05/91)
I have been trying to use the sort facility on machine (Sun 3/80, SunOS 4.0.3) for the past few days. I have run into several problems first the -z option gets maestrosh[277]sort -m -z 5000 breakup sort: invalid use of command line options maestrosh[278] Second to avoid running out of memory during the sort phase I have had to use the split function and then sort each file individually followed by a merge phase. When merging the result I have tried to use the -u option ie. sort -n -mu file1 file2 >! outputfile but this causes sort to include only one of the following lines -----line 1 (wrapped around to avoid news/mailer problems) 77 0 4408 4410 1 DOCUMENT 1 0 8 SECTION 1 7 9 HEADING 1 8 /DOCUMENT/SECTION/HEADING -----line 2 77 0 4689 4690 1 DOCUMENT 1 0 8 SECTION 2 15 9 HEADING 1 16 /DOCUMENT/SECTION/HEADING These lines are obviously different. In the short term I have gotten around the problem by leaving the uniqing until after the merge phase and using the uniq command which appears to work. As far as I can tell from the man page the sort fields should be the entire line unless I specify otherwise on the command line. Can anyone shed some light on these on these two problems? Mark Levison levisonm@qucis.queensu.ca
brnstnd@kramden.acf.nyu.edu (Dan Bernstein) (03/12/91)
In article <1105@maestro.queensu.CA> levisonm@qucis.queensu.CA (Mark Levison) writes: > maestrosh[277]sort -m -z 5000 breakup > sort: invalid use of command line options Some sorts insist that you leave out the space: -z5000. (Pretty silly interface, if you ask me. I don't see why sort should care what the maximum line length is beforehand.) > sort -n -mu file1 file2 >! outputfile Wasn't there some talk about this in comp.bugs.sys5 recently? Basically, yes, that's what some sorts do with -n, and you should either get used to it or use some field specifiers to avoid it. ---Dan