This option is extremely important when wget is used within cron scripts. the option not being available can lead to multiple stuck scripts that need to be cleaned by hand or blasted by some other shell script.
In 2008, Matthias Kaehlcke <matthias@kaehlcke.net> posted a patch to the mailing list for this: http://lists.busybox.net/pipermail/busybox/2008-August/066797.html It does not appear his patch was ever applied; there was some discussion about bloat and then Denys suggested: > To be honest, whole dance with alarm() in wget needs rewriting, > poll(..., seconds*1000) + read() would allow signal-less handling > of progressmeter and/or timeout. So doing that is probably the right next step here.
Created attachment 2329 [details] FEATURE_WGET_TIMEOUT: in wget applet, support "-T seconds" option for network read timeout. As Denys suggested, I reworked alarm() -> poll() for the progress meter, and then used that change to add the -T seconds timeout option for wget, which is now FEATURE_WGET_TIMEOUT in the attached patch. See more details on this mailing list post: http://lists.busybox.net/pipermail/busybox/2010-August/073074.html
Patch is applied to git, will be in 1.18.x Thanks!