infoscape

Screenshot.

Since some days living in my desktop, I’ve decided to release a screenshot of how infoscape 1.1 will look… the general appearance is almost the same but a side knob has been added to drag the window up or down (mouse wheel no longer used). Options also improved: headline repeat and channel polling interval can be now configured for each feed separately. More details when released :).

At least! infoscape 1.0 is out with lots of improvements. But this keeps being a proof-of-concept software and there are many pending/unreleased features awaiting to be in the wild. Coming soon, version 1.1 containing:

Well, while this comes, don’t miss v1.0 and the changelog included in the package.

Even when version 1.0 is not publicly ready, development and improvements are going on in experimental branch of the software, increasing software quality and features (and bug count) day by day. Current development tasks in this experimental branch are:

When are experimental improvements being integrated into “release” version? well, it depends on how fast the features get stable, this means that bugs and changes per week into the code related to these are very low or zero.

Since the begining of this project, I’ve been completely astonished about the amount of memory used by infoscape (between 20 & 30 Mb’s, according to my measurements).

After playing a little bit with Windows performance administrative tools I´ve found that memory usage grows in around 15 Mb’s when letting MSXML engine fetch the XML files from HTTP sources (passing directly the URL to IXMLDOMDocument::load()). This doesn’t happen when sources are located in my computer’s hard drive, and not the internet.

This probably happens because retrieving files through HTTP protocol ‘wakes up’ a lot of shared libraries to support the discovery and download of these files. A good workaround for this would be implementing my own light-weight HTTP file stream accessor, in order to avoid MSXML increasing infoscape memory consumption with a lot of libraries and dynamic structures which probably are not needed.

Anyway, this will be implemented (if possible) first in the “experimental” release of infoscape, and will not be ported to “public” release until the next version to 1.0.

A few hours ago the candidate source code for infoscape 1.0 was freezed, which means that no more changes/improvements will be intruduced to the software unless really critical failures arise. This is the first step to final version 1.0 delivery.

Why not release right now? well, I’m testing the software in many ways (memory/processor consumption during long-term runs, stress tests on user interface, etc.) and I´m also applying for a place inside download.com archives (and getting this takes time, trust me).

Previously published goals for infoscape 1.0 are completely fulfilled but under testing. Finally, bug solving is going beyond the configuration dialog and much more things are being optimized to improve application robustness.

Another line of development (which will be probably delivered in 1.0) covers text processing for headlines. Unfortunately, many news channels doesn’t stick very well to RSS and XML standards, inserting HTML tags and entities inside standard tags. Currently infoscape developer works hard in order to make his (our) software clean-up correctly the retrieved text.

But this is not an easy task, HTML standards covers lots of entities and symbols which not even advanced browsers are fully prepared too. Another fact is execution speed and resource usage… string processing and pattern matching for each displayed text takes time and one of the aims of infoscape is to keep the resource usage as low as possible, while keeping a user-friendly interface to the user.