Project Themis (original) (raw)

Back from our nap
Written by Raymond Rodgers on Sunday January 15, 2006 10:24:39 AM
You may consider this the first sign that the Themis developers have awakened. Check the dev mailing list for an update email... this web site... will be reworked and updated further in time.

Merry Christmas to All!
Written by Raymond Rodgers on Friday December 24, 2004 09:37:46 AM
Hi folks! I just want to take this time to wish all the developers on this project and all admirers and potential users of Themis a very merry Christmas! I know that we've been quite for several months now, but that doesn't mean this project is dead. Unfortunately, most of us developers have been dealing with real life a lot more than we would like to, and haven't been busy as a result. That said, I would like to point out that I silently added RSS support for the web site, so that anyone using an RSS reader can easily determine when we post a new article to this site. People using Firefox 1.0 (on Windows and Linux at least, possibly BeOS as well though) should have noticed addition nearly a month ago, however, here is the actual link for everyone else: http://themis.sourceforge.net/rss.php. That said, I again want to wish everyone a merry Christmas, and a happy new year!

Merry Christmas to All!

Written by Raymond Rodgers on Friday December 24, 2004 09:37:46 AM

Hi folks! I just want to take this time to wish all the developers on this project and all admirers and potential users of Themis a very merry Christmas! I know that we've been quite for several months now, but that doesn't mean this project is dead. Unfortunately, most of us developers have been dealing with real life a lot more than we would like to, and haven't been busy as a result. That said, I would like to point out that I silently added RSS support for the web site, so that anyone using an RSS reader can easily determine when we post a new article to this site. People using Firefox 1.0 (on Windows and Linux at least, possibly BeOS as well though) should have noticed addition nearly a month ago, however, here is the actual link for everyone else: http://themis.sourceforge.net/rss.php. That said, I again want to wish everyone a merry Christmas, and a happy new year!

A few words before BeGeistert.
Written by Raymond Rodgers on Tuesday October 5, 2004 03:55:18 PM
Hi Folks! This is just a little note to give everyone a heads up on the status of Themis just a little before BeGeistert. The team has been busy over the last couple of months refining a lot of Themis' rough edges, and improving performance. While I can't say that we'll be able to show a finished web browser, I believe that what we will be able to show is very exciting nonetheless. Over the last 60 days, we've seen improvements in the HTML parser and the beginnings of Cascading Style Sheet support courtesy of the hard work of Mark Hellegers. (Give him a huge pat on the back as he's in the process of finishing up his Master's degree!) Michael Weirauch has been busy with the redesign of the settings user interface as well as URL and web site management. I, myself, have been conducting the fourth major rewrite of the HTTP protocol support, and it looks to be the most complete and best yet. And Olivier Milla has been trifling around with a little thing called the renderer, and doing minor things such as displaying web content. Judge it for it's capabilities, not it's content. ;-) And to top it off, there have been many stability fixes, improved cookie support, inclusion of Cryptlib for SSL support, and a newer version of zlib to enable gzip content-encoded communications. So we're busy, even if we are quiet, and I hope that we'll be able to bring a smile to your face in a few days' time.Raymond

A few words before BeGeistert.

Written by Raymond Rodgers on Tuesday October 5, 2004 03:55:18 PM

Hi Folks! This is just a little note to give everyone a heads up on the status of Themis just a little before BeGeistert. The team has been busy over the last couple of months refining a lot of Themis' rough edges, and improving performance. While I can't say that we'll be able to show a finished web browser, I believe that what we will be able to show is very exciting nonetheless. Over the last 60 days, we've seen improvements in the HTML parser and the beginnings of Cascading Style Sheet support courtesy of the hard work of Mark Hellegers. (Give him a huge pat on the back as he's in the process of finishing up his Master's degree!) Michael Weirauch has been busy with the redesign of the settings user interface as well as URL and web site management. I, myself, have been conducting the fourth major rewrite of the HTTP protocol support, and it looks to be the most complete and best yet. And Olivier Milla has been trifling around with a little thing called the renderer, and doing minor things such as displaying web content. Judge it for it's capabilities, not it's content. ;-) And to top it off, there have been many stability fixes, improved cookie support, inclusion of Cryptlib for SSL support, and a newer version of zlib to enable gzip content-encoded communications. So we're busy, even if we are quiet, and I hope that we'll be able to bring a smile to your face in a few days' time.Raymond

Back on Track...
Written by Raymond Rodgers on Wednesday July 28, 2004 10:14:54 AM
After several months of scratching our heads, we have finally resolved a crashing bug that had been hindering development of the renderer. Frankly, the bug had been a major pain in the butt to find because it gave the symptoms of a stack corruption, and only occurred when the renderer add-on was placed in the add-on directory. This was not a happy situation. Reviewing the code in the renderer repeatedly revealed nothing. The message system which had been recently modified when the bug appeared was not the cause, though the crash did occur during the process of receiving messages. Needless to say we were stumped. Today, the fog was lifted. While looking for the cause of this bug, I added a few printf() calls to the renderer to try to narrow down the precise cause of the bug. I found nothing... initially. Everything seemed to be functioning as it should. I started playing around with different combinations of add-ons to verify that the crash only occurred with the renderer loaded... It was then that I realized something important. One of the last printf statements was only displayed if the cache system add-on had been loaded, and my output showed that it had been. The problem was, that I hadn't even built the cache system add-on...! What appeared to be happening was that the renderer add-on was being loaded, which (like any other Themis add-on) triggers a broadcast message notifying all add-ons and message system members that a new add-on was in memory. This message includes information about the plug-in which was being compared bitwise with a specific value to identify the cache plug-in. Suddenly the pieces were falling into place. If the value that the renderer was yielding matched the bitwise comparison for the cache, then the renderer's plug-in object was being mistakenly cast as the cache plug-in object type. That in turn would mean that something calling a function in mis-cast object would trigger a segmentation fault... Which is precisely what we were getting... I tested the theory by changing the checks for the cache system, and the theory panned out. The problem was fixed. And I felt like an idiot for an hour. This all could have been avoided with a bit more careful planning and understanding of bitwise operations; going back years rather than just months. I apologize for the confusion and the delays, but the important thing is we're back on track!

Back on Track...

Written by Raymond Rodgers on Wednesday July 28, 2004 10:14:54 AM

After several months of scratching our heads, we have finally resolved a crashing bug that had been hindering development of the renderer. Frankly, the bug had been a major pain in the butt to find because it gave the symptoms of a stack corruption, and only occurred when the renderer add-on was placed in the add-on directory. This was not a happy situation. Reviewing the code in the renderer repeatedly revealed nothing. The message system which had been recently modified when the bug appeared was not the cause, though the crash did occur during the process of receiving messages. Needless to say we were stumped. Today, the fog was lifted. While looking for the cause of this bug, I added a few printf() calls to the renderer to try to narrow down the precise cause of the bug. I found nothing... initially. Everything seemed to be functioning as it should. I started playing around with different combinations of add-ons to verify that the crash only occurred with the renderer loaded... It was then that I realized something important. One of the last printf statements was only displayed if the cache system add-on had been loaded, and my output showed that it had been. The problem was, that I hadn't even built the cache system add-on...! What appeared to be happening was that the renderer add-on was being loaded, which (like any other Themis add-on) triggers a broadcast message notifying all add-ons and message system members that a new add-on was in memory. This message includes information about the plug-in which was being compared bitwise with a specific value to identify the cache plug-in. Suddenly the pieces were falling into place. If the value that the renderer was yielding matched the bitwise comparison for the cache, then the renderer's plug-in object was being mistakenly cast as the cache plug-in object type. That in turn would mean that something calling a function in mis-cast object would trigger a segmentation fault... Which is precisely what we were getting... I tested the theory by changing the checks for the cache system, and the theory panned out. The problem was fixed. And I felt like an idiot for an hour. This all could have been avoided with a bit more careful planning and understanding of bitwise operations; going back years rather than just months. I apologize for the confusion and the delays, but the important thing is we're back on track!

One Step at a Time...
Written by Raymond Rodgers on Sunday February 29, 2004 03:25:03 PM
Hi folks,This is just a quick update to let the general populous know that we're still moving forward, even if we are a bit quiet right now. In recent weeks, we've seen the ability to render text appear, even if it does overlap itself due to a lack of positioning at this point, and some improvements in the Document Object Model, as well as cache system and user interface.We've also been asked, again, about using an existing rendering engine such as Gecko. I don't intend this to be in any way harsh, but this project will continue to pursue development of a new rendering engine. However, if there are individuals out there that wish to pursue adapting Themis to use an existing rendering engine, check the mailing list section at the SourceForge project page and join the themis-altre mailing list (which should be available in a few hours) to discus it and how to go about it. We likely won't actively code Themis to work with another rendering engine, but we're willing to help others that are interested in doing so.That'll do it for now, hope you're having a great 2004 so far!

One Step at a Time...

Written by Raymond Rodgers on Sunday February 29, 2004 03:25:03 PM

Hi folks,This is just a quick update to let the general populous know that we're still moving forward, even if we are a bit quiet right now. In recent weeks, we've seen the ability to render text appear, even if it does overlap itself due to a lack of positioning at this point, and some improvements in the Document Object Model, as well as cache system and user interface.We've also been asked, again, about using an existing rendering engine such as Gecko. I don't intend this to be in any way harsh, but this project will continue to pursue development of a new rendering engine. However, if there are individuals out there that wish to pursue adapting Themis to use an existing rendering engine, check the mailing list section at the SourceForge project page and join the themis-altre mailing list (which should be available in a few hours) to discus it and how to go about it. We likely won't actively code Themis to work with another rendering engine, but we're willing to help others that are interested in doing so.That'll do it for now, hope you're having a great 2004 so far!