Friday Sep 08, 2006

Technorati needs owl:sameAs

There are two urls to access this blog. As a result Techorati, which just is a feed aggregator search engine, thinks there are two blogs:

These two urls point in fact to the same blog, that have the same atom feed, yet the technorati listings for its two blogs are usually out of sync ( has not been updated for 115 days), and the link count of one and the other do not match up. Really they need some functionality so that one can claim that they are both owl:sameAs each other.

And also less annoying advertising!

Thursday Aug 18, 2005

Atom 1.0 is an IETF standard

The Atom Syndication format is now a proposed IETF standard, which means that it is a first class IETF citizen. An html version as well as links to the developing Atom protocol are available on the web site.

I am really happy to be listed among the contributors, and I think Atom is a great improvement over RSS2.0. I am not yet sure how much of an improvement it is over RSS1.0, as I don't know the RSS1.0 standard in enough detail to be able to comment. The main advantage that RSS1.0 still has over Atom is that, being built on RDF, it has a very clear and solid extensibility framework. (That RSS1.0 needed some improvement is in part visible by the fact that an RSS1.1 draft spec has been delivered recently) Atom could have become clearly and unequivocally a better standard than RSS1.0 had it been possible to tweak it just a little bit to make it RDF compatible. The case for the extensibility advantages of RDF were very clearly made by Dan Brickley in an email to the atom mailing list a year ago, on the 18 August 2004. But as the rest of the thread shows some people found it very difficult to understand this point. There is clearly a lot of learning by experience the hard way that is going to take place before Dan's points are well understood and assimilated.

What Atom has now clearly achieved though is to add clarity to every other aspect of syndication. Work on the RDF side of things is continuing on on the atom-owl mailing list which will hopefully help provide a standard ontology for the atom format.

And by the way, BlogEd (when publishing to a simple web server using ftp such as my personal blog), now publishes an Atom 1.0 feed.

Monday Aug 08, 2005

BlogEd 0.7rc3 is out

Before going on my cycle tour I quietly released a new version of BlogEd. The time zone problem has been fixed, so that one can now post to roller from all over the world. A few other bug fixes, but otherwise not much. There should be more fixes later in September when various groups come back from holidays.

Saturday Jul 02, 2005

A post in Honolulu Time Zone

I am sending this post from France (UTC/GMT +2hours). It was created at 12:54 pm (local time) on Saturday. I have set my new CVS version of BlogEd to think that the server interprets dates as being in the Pacific/Honolulu Time Zone, (UTC/GMT +10hours) where it was 0:54 am. In fact the server believes it is in the "America/Los Angeles" Time Zone where it was 3:54 am (no idea where it really is btw). As a result this post should appear to have been created 3 hours earlier than it in fact was! Let's test this.

Indeed it worked. Roller gave this post the following date ( Jul 02 2005, 12:54:02 AM PDT ).

(This will change in the future, as I will be wanting to move the time zone of BlogEd to correctly be the Pacific/Los Angeles time zone)

PS. I am starting to think that the xmlrpc protocol was invented by someone who only used a PC and did not understand much about the network.

PPS. What is going to happen if multiple servers get deployed in different Time Zones for fail over capabilities, all hiding behind the same url? Ouch!

PPPS. I just noticed that one can set the time zone in Roller in which one would like the dates to appear [Website Tab - User link]. I have just changed it to Paris time (CEST), as this will make it clearer to my readers what time of day it was when I wrote my entry. As a result my posts now appear dated as ( Jul 02 2005, 09:54:02 AM CEST ) which is indeed still 3 hours ahead of time.

PPPPS. I have now changed BlogEd to generate times in US Pacific time, and now the time stamp for my entry appears correctly as ( Jul 02 2005, 12:54:02 PM CEST ). So changing the time zone displayed does not yet change Rollers interpretation of the time I am sending to it. I still need to know what time zone the server thinks it is in.

Thursday Jun 30, 2005

Time Zones and XML-RPC

I have had to hack the Apache xmlrpc library to add support for specifying the TimeZone of the remote server. I'll post the patch there as soon as I am happy with it.

I have to say though that I am very surprised that code that has been out there for such a long time has not had such a patch applied to it yet. Of course that something so hyped as XML-RPC has such a fundamental deficiency - that it forces dates to be sent without the time zone information - is itself mind boggling. How could anyone invent a protocol for remote communication and screw up dates so badly? Essentially this means that unless you know what time zone the server is in or you live in the same time zone as the server, all dates exchanges are meaningless. To illustrate: imagine you are speaking to someone on the phone and ask them what time it is. They tell you "8 o'clock". If you don't know where they are in the world that answer will simply be meaningless. Well that is what xml-rpc forces one to do! Would it have been difficult to add time zone information? Not at all. 4 characters more in an xml file.

Friday Jun 24, 2005

BlogEd 0.7rc2 is out!

Release Candidate 2 of BlogEd 0.7 is out!

For more information on how to obtain it, see the BlogEd site. You can publish to Roller by using the properties described in an earlier post on this blog.  

There are many improvements since the last release.

  • James Gosling has added some very helpful options to enable the author to tune the appearance of the blogs on the Roller site. In the Preferences menu, one can now choose to have the illustration appear on the right or the left of the post, or keep it as previously inside a table.
  • For MetaWeblog enabled servers that don't allow the client to specify the creation date of the post, one can ask BlogEd to add this date at the end of each post. This will help readers understand the sequence of the post and keep the chronology consistent. The Roller on and on Sun's internal blog have just been updated to understand the non standard pubDate field both for new entries and entry updates.
  • James Gosling also added a button in the preferences menu to allow one to republish the whole blog. This is important if you decide to make a change to the display of your posts. Use with care. If it take a lot too long to update your blog, this may be the reason of your problem
  • Many bug fixes that were pointed out since the last release
One nice feature that falls out of the redesign of BlogEd is that it is possible to publish your Blog in two different places at once, as illustrated by original and the new Roller version of James Gosling's "On the Java Road..." blog.

Try it out!

Mmh. Is there a time delay between the publishing of a new entry and its appearance on the Roller site? Indeed there is: xmlrpc on which the MetaWeblog API is founded does not specify the time zones when sending dates across the internet! So if you are using BlogEd to send posts to an instance of Roller, make sure you are in the same time zone or to the West of the server you are sending posts to (until we find a way to fix this problem).

Sunday May 29, 2005

final beta of BlogEd 0.7 ?

I have now released what I hope to be the final beta of what I was thinking of calling BlogEd 0.7. It is at the same location as the initial release. Don't forget to clear your jnlp cache before testing it.


  • supports categories now
  • fixes bugs pointed out to me in the previous release
  • properties panel now tests the xml-rpc connection
  • all the remote entries are downloaded in one go on first setup
  • clean up html before sending and receiving it
 I think this is quite usable now. I may have encountered a major annoyance with the kafenio html editor component. It seems to freeze when some non ascii characters are typed in. I don't yet know how deep this problem is. Hopefully I will be able to work out the cause with the Kafenio team.

Aside from that unknown this should be good enough for 0.7. I will try to implement the following features before the release:

  • an About panel that thanks all the open source software teams who have developed libraries on which BlogEd relies
  • spelling mistakes and other UI nits that can easily be fixed
  • better support for working with proxies
  • secure https connections?
 Ah, well... The more I think about what to add, the more come to mind... Best if you let me know what you think

Friday May 27, 2005

My first category post

This should end up in the BlogEd category. I posted this with BlogEd. And it seems to work correctly! Cool. I was even able to move a post to one category and then back.

Monday May 23, 2005

new beta of bloged

I have just placed a new beta of bloged in the same location as the previous beta release. This is an intermediary beta release. There are a lot of improvements in the code, not all of them visible to the end user.

There was a lot of work in the look and feel side of things. Not all of it is quite finished, but it should give an idea of where things are moving to.  
  • The infinite progress panel is in
  • setting the properties is much more userfriendly, though there are a few things that still can be done
  • error in xml-rpc comunication are much clearer
  • the general UI is cleaner. Is this better? I have lots of ideas on improving things a lot more, but there is just one of me...
  • the user is notified that editing the blog header is not helpful in the MetaWeblog api publishing side of things

Some things I did not get to put in, but which I will be working on in the next week:  I'd be glad to hear of any intial thoughts, keeping in mind that there are some things that still need to be done. Are people happy with the general direction this is evolving in?
  • categories functionality
  • improving the properties panel, so that the user can choose to download all remote entries when setting up his connection. Also I want to do a real xml-rpc test.
  • fix some of the smaller unclarities of language people pointed out in the previous review
  • speed up of initial synchronisation

You may need to clear your jnlp cache to get it to download the new version. Is there a way to make sure it picks up on new versions?

Note: before you start testing this version you should delete (or displace) your old directory that you created previously. This will give you a fresh start.

Working around MetaWeblog xml-rpc vagueness

The MetaWeblog API is an xml-rpc communication mechanism which I am using to post these messages from BlogEd. It is a little like Java Remote Method invocation, except that it uses xml to serialise the object states. Xml-rpc is allready very vague and imprecise, but the MetaWeblog API does not help. There are for example no agreed upon error messages as there are in HTTP. So that it is not possible to tell with certainty if an error message to a request to update a remote blog entry is telling you that
  • the password used to log in is wrong
  • the remote entry no longer exists
  • the remote entry never existed
  • some other error in the communication occurred

There is a little bit of explanatory text that will help the end user if he speaks the preferred language of the server author. In order to help BlogEd users understand the predicament I am in as a programmer I have given them all the information they need to make up their mind. Where it makes sense to give the user a choice, I pop up a dialog where they can inspect the xml-rpc messages themselves.

This makes me feel less lonely. In any case the user will probably know that he deleted an entry using the web interface a few days before.

Monday May 16, 2005

some UI changes in the pipeline

I have been working during the past week on incorporating some of the suggestions I received from my friendly beta testers. Some of this has required me to do a lot more work with Swing than I ever did before, so the learning curve has been a little steep, but I think I am really getting the hang of it now, though there is clearly still a huge amount out there :-)

One element I am hoping to make a little clearer is that the first entry in BlogEd is the title entry. I am doing this by placing information about the entry into the left hand pane. Also since the title entry cannot be published using the MetaWeblog API I have added a little glass pane with a message to indicate this.

The categories combo box would only be visible for MetaWeblog API users at first. This still needs to be implemented.

The look is a lot better I think already. Even though it may not seem like a huge change, when one is learning to Swing, getting a few small things right can be quite tricky. What is nice though is to see how mature the framework now is. So at least I don't have to worry as I did in Java1.0 days about whether its a knowledge or a framework problem. Now I know that it's just my problem. The documentation for Swing is also really good now.

And the more I play with this the more ideas I am having...

Oh. I have not yet put the above in CVS... I hope to have a new beta this week.

Tuesday May 03, 2005

BlogEd Beta release

This is my first BlogEd entry on my external Sun Blog.

I am playing with a new beta release that I have just made availabe

To get this to work with this version of Roller you need to open up the "Edit > Blog Properties" menu and choose the blog info pane and set the "MetaWeblog API" publication method. Here are the fields I set in the Blog Properties pane:

Property Value
service url
blog id bblfish
user id bblfish (same as above)
password won't tell

When you next publish an entry it will first download as many remote entries as Roller will allow, and then publish the new ones you have written. As you should see, BlogEd allows one to add a little picture next to your text to illustrate your post.

This is still a beta release, and prone to change. But it is stabilising quite well.

One thing we still need to work on is some way of securing the xml-rpc communication between the client and the server.




« June 2016