Older blog entries for mpesenti (starting at number 7)

20 Jan 2004 (updated 20 Jan 2004 at 10:58 UTC) »

I've been forced to stay mostly out of the loop for the last two months to be able to complete my last univ project. Hopefully I'll get out of there soon ;) I'm grateful to Christian that has been doing a great work maintaining epiphany in this period.

I finally managed to get some time to look at a solution for the problems in new downloads design. I think we reached a good compromise and we have plenty of space for improving in next releases. We need to fix the rhough edeges of the implementation ASAP. Feedback on usability list was useful, especially Alan and Clarck feedback (thanks guys). I'll certainly look at implementing more of Alan suggestions in the future, they are very interesting.

Bye Mark!

The first memory I have of Ettore is a chat with him and Chema. I like to think to them that way, still together. I was in Milan today and everything remember me of him: talking with other gnomers, looking at bug mail and thinking of his bugs I havent fixed, discussing the downloader and remembering his mail about it ... It hurts. My heart and my thoughts are with his family and with Ximians. Keep up guys, for him !

Wow long time not posting here. I have to try out John Perry Structured procrastination theory.

I was looking at Epiphany 1.2 plan, in particular to the "Aspects to improve" section.

Better communication with GNOME teams. Good. I like the fact that we have been able to provide feedback on menu/toolbars/combo gtk new widgets. GTK people are being very very open about suggestions/bug reports and we really appreciate it. (Btw I'm impressed by the improvements made to gtk 2.4, it's taking some more time than planned, but it's damn worth).

Better communication with testers. Good. I've been surprised by the good feedback we got for 1.0, after the gnomedesktop flames. We are getting a lot of bug reports and productive feedback on mailing lists. (We are still at 200 bugs. Yay ! Christian is doing a great work.)

I think we can do a better work avoiding regressions. Not too bad, we had a few regression but it's acceptable given we need to concentrate adding/reworking fetaures now.

Connection with Mozilla people. Improving, we got in some fixes we needed pretty fastly (Christian established the new record, only 1 day !). Though some really bad, old, embedding bugs are still there.

Design should not block on me. This is where I'm not satisfied. I tried to get people more involved in the design of new features/improvements, but the situation is still exactly the same as 1.0 development, things block until I have the time to make them moving. This slow down development quite a bit, because it blocks possible contributors to code.

The alternative is to make interface changes without thinking to them much, hoping they work and trying to fix problems when they show. But I dont think I like the idea.

The worst is that I dont quite understand why it's so hard to get contributions in this field. Maybe lack of interest (I guess most prefer code to human interface design) or lack of a documented methodology. Maybe a bit of both. It's really frustrating for a maintainer.

In practice. In practice I'm generally happy on how we have been progressing. I'm keeping track of the work done in the Plan itself, deleting items we completed.


I have been not able to do any work on Storage for a few weeks. Luckily I see Seth is rocking on natural language. I must be back on it soon !

We fixed/closed several bugs today and we are finally back under 200 (enanchements included). It would be good if we could keep them at this level but ... that will be hard ;) I found an uninitialized variable that could cause the "reappearing location entry" bug. I really hope it's that, the bug is hard to reproduce (impossible for me recently).

Email viruses are getting me crazy. I get like 7-8 message of 100 Kb per hour now, I'm sure soon my mailbox will start to get full again during the night ... Sucks.

I'm trying to hack an html importer for storage. It should walk the DOM tree and extract informations like title, links, keywords ... Open Text Summarizer is going to be useful. I want to solve the authentication issue in libstorage-translators before though.

And finally I remember to say ... gnome-blog rocks !

George posted an interesting plan about desktop lockdown. We have been discussing something similar for ephy a few days ago. A point that I still dont have very clear is how to disable access to menus functions. I'm not sure if it's better to allow admins to use a custom xml definition (which is quite simple with GtkUIManager and includes only layout information, unlike bonobo), or to have boolean gconf keys to disable menu items (or rather group of items).

I think we have failed to communicate how Epiphany is an evolution of Galeon and not a replacement. I had a few interviews in italian about 1.0 release and everyone asked something like "Why it has been necessary to replace Galeon with Epiphany ?". The e-week article seem to show the same misunderstanding ... I guess the root of miscommunication has been my decision to leave galeon development after the bigger steps in the "epiphany direction" was made: that suggest I wanted something radically different from galeon2, while I was mainly concerned to not waste the improvements made in the 1.3 development cycle. Anyway I'm happy of how the Epiphany/Galeon mess is evolving: the targets of the two projects (and codebases) start to differ more strongly and the flamewars seem to be over. I feel much more productive with the new development team, and I'm sure the same is true for them. The feedback on 1.0 has been generally very good, which surprised me after the heated reactions on gnomedesktop. Last here with the endless discussions on galeon-devel, the fork, the GNOME inclusion with related flames, has been very stressing for me, maybe that time is finally over ... A lot of people with their support and work, convinced me that it was worth to keep up after the strong delusion of the fork, thanks guys.

Christian rewrote the xml loading code for nodes in head, with some performance and code cleanliness gains. Yesterday has been a good day on bugzilla front. We closed several bugs and reporters finally calmed down a bit :) The fullscreen interface changes are shaping out in my head a bit more, I think I'm going to make them soon. I wonder if someone will ever get to review my focus patch for gtkmozembed, that's the worst bug we have so far ... Even if the patch is not correct (I think it is and people are using it without problems, but mozilla focus is a mistery), it's sad that something so basic has not been fixed in 3 years :/

16 Sep 2003 (updated 16 Sep 2003 at 22:14 UTC) »

We are working for a bugfix release on the stable branch to go with mozilla 1.4.1. Christian did a great work closing a lot of dups in bugzilla and posting fix for them. I have still to review his gnome print patch though ... bad me.

I finally got my work on StorageStore in a working state. Waiting for seth approval to check it in completely. It seem to work nearly out of the box with redhat 9 installation, which will be good for testing.

Had to explain someone how to organize his photos yesterday (on windows). It was a real pain, I hardly managed to explain him how to put them from photocamera virtual drive on the images folder, but I had to give up on organizing them in subfolders part ... Nautilus guys are trying out a new UI that will include a spatial model as default. I think it's an interesting development. While I agree with Norman's critics to a purely spatial organization of data (that's why I actually prefer the "object model" definition), I think the navigational model introduce an extra level of abstraction to try to work around problems on the way data is accessed. You gain some speed moving through the directories, but the problem isnt really solved: when working on projects with several files involved, I usually end up spending more time navigating in folders, than doing the actual work. Going back to a more natural (less abstracted) conceptual model could encourage the development of more natural and efficient ways for accessing/filtering data. Basically my feeling nautilus developers are doing a great step in the right direction, good luck :)

In the last few days I've not get much done in epiphany. I've been busy with 1.0 interviews, and getting my system back to work in a decent way: it's hard to keep a working desktop when hacking on unstable stuff.

Though I think epiphany unstable branch is progressing quite well. We ported our menu stuff to gtk 2.3 and solved several suckiness in the egg based implementation; Mathias is doing a great work fixing bugs and improving gtkuimanager api. I'm also very happy with the progresses we made on font configuration, it finally start to make sense :) While I've been idle Christian produced a few patches I need to look at, it sucks to leave unreviewed stuff in bugzilla. Also someone volountered to work on kiosk support, and posted a list of necessary features, rock ! The negative side of 1.0 is that we are getting lot of work to do in bugzilla, hopefully I'll be able to get someone to help with it. Volounters ?;)

On the storage front, I've been working to split libstorage implementation in StorageStore and StorageItem, the main advantage being the ability to use a local configuration of PostGre without requiring user and password. I think I'm not far from something working, but my borked system stopped me to get it finished. Seth posted a nice plan on things that need to get done on the long term. It's a lot of stuff, but I think we can gradually get there :)

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!