I have so many open windows I didn't press "send" on. Trying to flush 'em...
On 9/17/19 6:38 PM, enh wrote: > ping? (you submitted the test, but not the fix :-) ) Sorry, instead of: $ diffstat ../0001-tar-fix-creation-of-tar-files-containing-sparse-file.patch tar.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-) I went with: $ git show | diffstat tar.c | 17 +++++------------ 1 file changed, 5 insertions(+), 12 deletions(-) Needed a bit more brain than I had at the time to work through it. Rob P.S. The TT.sparse allocation is retained and realloc()ed to the appropriate size next time, freeing and reallocating it doesn't really help? It's hardly ever a page. Does using a local variable instead of incrementing the final variable each time produce noticeably smaller code? I'd think the optimizer could notice that, it's doing so much else ridiculous lifetime analysis these days you need to stick barriers in stuff to _stop_ it. And declaring another i when there's already an i in the enclosing context is uncomfortable, and it was free to use anyway between determining link type on line 267 and iterating through the extent overflow blocks on on 349. P.P.S. I _really_ should have called this project dorodango. _______________________________________________ Toybox mailing list [email protected] http://lists.landley.net/listinfo.cgi/toybox-landley.net
