Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ucsfcgl!cca.ucsf.edu!wet!epsilon From: epsilon@wet.UUCP (Eric P. Scott) Newsgroups: comp.sys.next Subject: Re: Usenet software Summary: Yes, yes, I'll release it Message-ID: <658@wet.UUCP> Date: 11 Oct 89 10:18:42 GMT References: <651@wet.UUCP> <539@cpdaux.UUCP> Reply-To: epsilon@wet.UUCP (Eric P. Scott) Distribution: usa Organization: Wetware Diversions, San Francisco Lines: 79 In article <539@cpdaux.UUCP> cpdaux!steve@apple.com (Steve Lemke) writes: >How's the performance on unpacking and expiring news with B News? I've >heard that cnews is much, much faster, but I had some difficulty in >trying to get it going. My other problem is that the modem isn't here >yet, so I'm not getting any news batches so as to test what I've compiled. My experience with B News is limited to this machine (wet) under SysV/386 and the NeXT. On the 386, the limiting factor on unpacking news is the speed of compress. B expire runs just fine with about 110MB in the news spool. (note that I'm using dbz on the Sys V machine). On the NeXT I'm NNTP-only, so I don't have to deal with compression. C expire may be a win, I don't know yet. >}Larry Wall's Configure has some problems. >}It doesn't like the GNU cpp, so it had to be faked out somewhat. > >I'm assuming you meant where it checks "abc=xyz" and you have to insert a >space before the '=' in order to get it pass the test? Yes. >I ran into several problems with the configure scripts also - I wonder >if the Configure script will be changed to be more NeXT-friendly. Well... I just brought up rrn under SunOs and it didn't do too much better. SunOS looks like both USG and BSD, and Configure happily brought in both ioctl and termio. And got other things wrong too. >}I added window size support to rn, It turns out that this is also in one of the rrn patches, and I ended up with doubled code (heh heh). >}It's going to undergo a bit more testing before I consider it >}ready for release, but the problems I've encountered seem not to >}be NeXT-dependent. > >Any chance when you're finished of sending the problems back to the >appropriate people so future relases might (gasp) be able to be >compiled with no problems by neophytes like myself instead of having >to hack through all this stuff? Who are the "appropriate people"? I've never been able to get a non-automated answer from Larry Wall. The supposed "rn 41" I got turned out to be identical to patchlevel 40, and the latest I could find at JPL was patchlevel 39. The best I can do is put together a patch archive and post it to news.software.b asking for merges. >What did you do about Pnews not being able to create a .newsrc? From >what I've seen, I don't think it's a NeXT-specific problem - something >about sed not being able to handle more than 10 something-or-others at >a time. Is there a known fix to this? Pnews doesn't create .newsrc; newsetup does, and that's invoked by rn. newsetup claims a limit of 10 w's per sed, but it's really 9. Basically you nuke all the "net" "mod" "fa" stuff that's no longer used and things are ok. This too is fixed in the rrn patches, although I think that misses putting "to" groups in .newsrc. >}A few more things too, that I don't remember at the moment. Like linking with the shareable library! Big win, especially on 8MB cubes. >We only have a single 660 mb drive, but it's only got about 260mb on it >right now. Any parameter changes we should make when we get a full news >feed going on it? (The machine has 8mb of RAM though we'll probably try >to bump that up to 16mb if it will help). Do you have a single partition or two? (If you're a netboot server the second partition is spoken for). Inodes are good. usenet likes many, many small files. stan@cs.sfsu.edu built the partition; I don't have the numbers handy. -=EPS=-