kay@warwick.UUCP (Kay Dekker) (07/07/86)
Here's the Path line from the header: Path: warwick!ukc!mcvax!seismo!harvard!think!nike!ucbcad!ucbvax!hplabs!tektronix!teklds!midas!jeffw And the offending section: >Oh, and take a look at the chord progression. Nothing in particular at first, >just remember it. Now look down to where at seees tg be `eadafg tgwardC eafgr assueafg qgur cgpq as af F eafgr!$ bust past w`ere t`ere's a D6'5&Wgw! _hi|(haxxmnml? It's as if the singer suddenly realized that magic was >mentioned - and the key drifts off in a totally different and wonderfully >rich direction. (Note to academics - cleverly done by reinterpreting the D >as a leading tone, nicht wahr?) If the song is a gem, this must be one of >the sparkles. Point 2: The unexpected happens naturally, gorgeously, before >you even know what's going on. This is the third article about this kind of mangle I've posted this year; OK, I may be NewsAdmin here, but I don't get to proofread *every* article that flows through warwick; how many articles are getting partially trashed in this way? - it's less obvious than the line-eater bug (of ill fame) cos the dross may not appear until quite a way through an article... If this *is* a compress bug (as someone who sounded knowledgeable guessed - I can't remember who...), then any sites that use compress for batching news may be vulnerable: maybe sites along the afflicted Path should state their version number & any local hacks. We might be able to kill this one off quickly. FYI, warwick is running 3.0, no hacks. Are there any later versions? Alternatively, it could be a bit (or bits) trashed during transfer of a bach that compressed OK; but surely all transmission protocols specify some sort of checksum/error recovery method? no? Distressed, Kay.
fair@ucbarpa.Berkeley.EDU.UUCP (07/12/86)
Kay (and anyone else trying to get diagnosis of network problems from the network community), It would help a great deal if you included the entire header of the article in question in your query about this problem. ucbvax keeps netnews online for 30 days, and it would take quite a while to guess which article you refer to from just the path line. For reference in solving the problem at hand, we use compression to send *to* `hplabs,' but I don't know off hand if they compress to us. We do not compress to `ucbcad' since they are on our local ethernet, and it costs us nothing to send to them except processor time. If you can provide the message-id of the mangled article, I can examine it on ucbvax, and ucbcad. keeper of the network news for ucbvax, and guardian of the gateway, Erik E. Fair ucbvax!fair fair@ucbarpa.berkeley.edu
george@sysvis.UUCP (07/15/86)
> If this *is* a compress bug (as someone who sounded knowledgeable guessed - > I can't remember who...), then any sites that use compress for batching news > may be vulnerable: maybe sites along the afflicted Path should state their > version number & any local hacks. We might be able to kill this one off > quickly. FYI, warwick is running 3.0, no hacks. Are there any later > versions? > Alternatively, it could be a bit (or bits) trashed during transfer of a > batch that compressed OK; but surely all transmission protocols specify some > sort of checksum/error recovery method? no? Maybe it was introduced by some wayward program called "junker" (as discussed so recently on this network).