http://radio.weblogs.com/0114065/categories/myProfession/

Friday, April 30, 2004 RE: Battelle on Google’s S-1: http://radio.weblogs.com/0114065/categories/myProfession/2004/04/30.html#a481

liquid labour day

1. paths /data/www/virtual/netfrag/sites/public/rssreader /home/service/bin

2. urls http://www.netfrag.org/rss/2.0/nfo.yakka.rdf http://www.netfrag.org/rss/1.0/nfo.yakka.rdf

doesn't work: 6 Mar 2003 22:57:40 GMT 13 Feb 2003 09:04:19 GMT

works: Tue, 11 Mar 2003 01:01:02 +0100

example pages/links for RssReader^?

error (no patchsets seem to be generated) /usr/local/bin/xcvs -c /etc/xcvs.conf ahhh, it's /etc/xcvs.config

rss 1.0: "rss-merge.xsl" requires "<item"s having an "about" attribute!!! it seems to get used as some id/key:

[...]

[...]

[...]

=> editing rss2rdf.xsl accordingly we don't have an url, so we use the title

Q: Better use cvs2rss??? Check it out!

build email-obfuscators into web-scripts (e.g. newsportal: "From: " header)

integrate "left" and "right" (or previous/next?) links into newsportal's "article.php"

flow: cvs commit -> cvsspam -> Mail::Dispatch::Gateway::News -> innd

add long versions of commands to "shortcuts::files" (e.g. f2s -> file2string) write module shortcuts::files::php|perl|basic|pascal which mimics file handling semantics of these languages/platforms

describe .symlinkrc

news-items which were converted to rss-items: where do we get additional (esp. "link") metadata from? could a parser for nfo.links.* be feasable? yup - should be: parse from e.g. "Content-Location:" or "Content-Base:" if MIME message no! don't use them: unfortunately they appear to have random blanks in them (where from? how does this?) e.g.: Content-Base: "http://support.lis.uiuc.edu/documentat ion/systems/RAID--Initial_Partition ing.html" Content-Location: "http://support.lis.uiuc.edu/documentat ion/systems/RAID--Initial_Partition ing.html"

maybe use text from the first part of the mime-message: http://support.lis.uiuc.edu/documentation/systems/RAID--Initial_Partitioning.html

or we have to assume "http://" (or parse the protocol from Content-Base/Location) and append the filename from "Content-Disposition", e.g.:

Content-Disposition: inline; filename="support.lis.uiuc.edu/documentation/systems/RAID--Initial_Partitioning.html"

... which seems contiguous.

ftp://ftp.suse.com/pub/suse/i386/9.1/boot/boot.iso

lamb - heaven - six feet under soundtrack

daemon - turns other processes into daemons fastjar - Jar creation utility java-common - Base of all Java packages java2-common - Common facilities for all Java2 environments jython - Python seamlessly integrated with Java jython-doc - Jython documentation including API docs sablevm - Free implementation of Java Virtual Machine (JVM) second edition kaffe - A JVM to run Java bytecode kaffe-common - Files shared between all Kaffe VM versions

full list: nfo.caesar.rdf nfo.dev.perl.rdf nfo.dev.php.rdf nfo.dev.python.rdf nfo.dev.ruby.rdf nfo.dev.tech.rdf nfo.faq.users.rdf nfo.links.computing.rdf nfo.links.misc.rdf nfo.log.cvs.rdf nfo.sysadmin.debian.rdf nfo.yakka.rdf site.rdf

first test: nfo.yakka nfo.dev.python.rdf site.rdf nfo.log.cvs.rdf

It turns out that the RSS feeds news2rss generates are RSS 2.0 ones. TWiki produces RDF 1.0. The rss-merge XSLT 1.1 stylesheet from http://ex-code.com/rss-merge/ however operates on RDF 1.0 as well. Solution? Find something to convert RSS 2.0 to RDF 1.0 or take completely different approach with RSS feed merging.

• rss2rdf.xsl:
  http://w3future.com/weblog/gems/rss2rdf.xsl
  http://www.peerfear.org/cgi-bin/viewcvs.cgi/newsmonster/xslt/rss2rdf.xsl?rev=1.29&content-type=text/vnd.viewcvs-markup

• rss2rdf.pl:
  http://www.pipetree.com/~dj/cgi-bin/rss2rdf.pl?code=yes
  XML::RSS

• rss2rss

* url input for http://netfrag.org/rssreader/

top-snapshots:

saxon: 10090 service 8 0 3896 3896 968 R 10.8 2.0 0:00 sablevm 10090 service 12 0 8732 8732 968 R 16.7 4.5 0:00 sablevm 10090 service 8 0 19352 18M 972 R 10.9 10.0 0:01 sablevm 10091 service 8 0 18160 17M 972 R 47.4 9.3 0:01 sablevm 10091 service 10 0 11928 11M 972 R 20.5 6.1 0:02 sablevm 10091 service 8 0 21328 20M 972 R 79.7 11.0 0:04 sablevm 10091 service 12 0 30620 29M 972 R 0.4 15.8 0:08 sablevm 10091 service 18 0 24232 23M 972 R 68.0 12.5 0:10 sablevm 10091 service 18 0 31160 30M 972 R 88.7 16.1 0:17 sablevm 10095 service 14 0 24064 23M 972 R 82.3 12.4 0:04 sablevm 10096 service 10 0 28804 28M 972 R 88.7 14.8 0:11 sablevm

sablotron:

libxslt/xsltproc: 10033 service 14 0 30688 29M 1136 R 37.1 15.8 0:02 xsltproc 10033 service 8 0 48528 47M 1136 R 11.9 25.0 0:03 xsltproc 10033 service 14 0 59852 58M 1136 R 5.2 30.9 0:04 xsltproc 10033 service 16 0 73012 71M 1136 R 3.8 37.7 0:04 xsltproc

times: #> time ./rss2rdf nfo.yakka nfo.dev.perl nfo.faq.users nfo.links.computing nfo.links.misc

saxon: real 3m20.816s user 0m43.577s sys 0m2.385s

sablotron: real 23m40.927s user 0m13.019s sys 0m7.346s 2x ./rss2rdf: line 1: 10165 Segmentation fault $sabcmd $2 $1 $3 param-rss-max-items=30 1x ./rss2rdf: line 1: 10246 Killed $sabcmd $2 $1 $3 param-rss-max-items=30

libxslt/xsltproc: real 3m27.296s user 0m8.769s sys 0m1.346s

Unfortunately "time --format=..." didn't work....

It appears that large input files make the xslt-processing (regardless of engine) spin around. So lets cut these files....

• check out why these are that large
• strip content/description to (e.g.) 200 chars max.
• use some timestamping in rss2rdf: timestamp=`date +%Y%m%d_%H%M_%S`

(Possible) Solutions:

• Use some limitations when generating RSS items from news items (news2rss):
  Look at the schema for RSS 0.91.