Re: Weird pkg_info behavior?
On 2008/10/5, Philip Guenther <[EMAIL PROTECTED]> wrote: > [In the context of pkg_info only downloading the first part of a > package to get the info] > > On Sat, Oct 4, 2008 at 5:09 AM, Slim Joe <[EMAIL PROTECTED]> wrote: >> >> Also I think it would be nice if the packages are dumped >> not into the present directory but in the $PKG_CACHE that >> I have set. > > pkg_add does that (assuming you actually export PKG_CACHE), > but I don't see why pkg_info would, given that it expects > to only download the start of the package file and not the > entire thing. Leaving partial package files in your cache > would be confusing, IMO. > > (Or did I misunderstand your suggestion?) Does the pkg install mechanism allow for resumable downloads? Up to now I've managed to finish all my installs in one run. But I assume resuming is possible, since pkg_add uses plain ftp to download the package tgz. It would then save bandwidth if pkg_add resumes from the partial download already made by pkg_info.
Re: Weird pkg_info behavior?
On 2008/10/4, Marc Espie <[EMAIL PROTECTED]> wrote: > pkg_info will only download the beginning of the package, since it only > needs the packing information, and we're very careful to store it at > the beginning. > > Now, a lot of FTP servers tend to not like abort in the middle of transfer, > especially when coupled with idiot firewalls. http transfers should be > slightly better. I'm not sure but the partial transfer appears to work with single (versioned) files. I think I'm using (I'm not at the machine atm) PKG_PATH=ftp://ftp.eu.openbsd.org/pub/OpenBSD/. Also I think it would be nice if the packages are dumped not into the present directory but in the $PKG_CACHE that I have set. This makes it convenient to export PKG_PATH=$PKG_CACHE:etc > Strangely enough, pkg transfers over scp is the fastest, due to using > a specialized protocol that avoids this set-up/tear-down altogether. > > So what you see is very much set-up/tear-down of connections, pkg_info > actually uses very little bandwidth, but it has a high latency. I'm not familiar with scp (never used it). Where's the list of scp mirrors?
Re: Weird pkg_info behavior?
On 2008/10/1, Stuart Henderson <[EMAIL PROTECTED]> wrote: > On 2008-10-01, Nick Guenther <[EMAIL PROTECTED]> wrote: >> If you are looking for package descriptions, install the ports tree >> and read the Makefiles. A lynx dump of http://www.openbsd.org/4.3_packages/i386.html seems more handy. > For 4.4/-current, landry@ has written a curses-based package browser, > pkg_mgr. It's in the ports tree and of course a package is available, > "pkg_add pkg_mgr". That's something to wait for then. From what I read it references openports.se.
Weird pkg_info behavior?
When I invoke something like "pkg_info vim*", pkg_info insists on downloading all the packages named "vim*". That is, I see a bunch of "vim*" packages on "." (present directory). Is there a way to get package info for a file not already downloaded or installed without such heavy bandwidth (just the package info). Note that my $PKG_PATH is set to ftp://[mirror site].
Tcsh does not recognize alt keys as meta
I installed tcsh using pkg_add on OBSD i386 4.3. The problem is that tcsh doesn't recognize either left or right alt key as meta. Instead I get different accented characters such as: (alt-backspace), f (alt-f), b (alt-b). This is true for both BSD and XTerm consoles. The alt keys work fine as meta under the default ksh shell. Out of the box, zsh also doesn't recognize the alt keys. However, adding "bindkey -m" to ~/.zshrc appears to "fix" the problem. Unfortunately, tcsh's bindkey doesn't have this "-m" option. Can somebody suggest a fix for enabling the console to recognize the alt keys as meta keys? I need to use tcsh because this is the shell I use in OSX and GNU/Linux. Thanks
