Dev day notes

Discussion of general topics about Seamonkey
Locked
User avatar
grayrest
Posts: 468
Joined: November 5th, 2002, 8:49 am
Location: Tribus!
Contact:

Dev day notes

Post by grayrest »

I don't have a cool blog site set up like other people, so here's my notes from the first half of dev day. Not everything is covered in detail because I expect people will post their slides, I've tried to only write about stuff that's not in the slides or is significant.

Code: Select all

Mozilla Developer Day 2004
``````````````````````````
choffman: Not much to say, mentioned that moz.org sold $15k in tshirts for (linuxcon/oscon/devday), good for them.

mitchell:
 - not much that her blog entry didn't say
 - desire to continue contributting to standards development (plugin/whatwg)
 - Mentioned increasing corporate interest/support
 - increasing web traffic, cracks top 1700 on release days

mshaver/beich:
 - shaver - aboout - heh
 - 2.0  goals:
    - Graphics toolkit to improve
        - Decent in xp capabilities before
        - Looking to improve, mentioned svg
        - Increased grapbics capabilities in PLATFORM not WEB
    - better embedded (mentions mozgtk/epiphany/mono)
        - express a desire to be part of a larger platform
    - Standalone apps are difficult
        - The boilerplate needs to be in place
        - want a toolkit that doesn't require so much to figure out
    - dev-test.m.o
        - they say that dev.mo.org is going to happen
    - desire to be able to
    - Q42004
        - replacement for gfx with cairo
            - getting close, not ready yet but expected to be
            - licensing issues
            - lot of momentum behind cairo
        - Dewipification
        - xtf landed
            - way of implementing tags in native code (more to come)
        - (some) svg
            -
        - xulrunner
            - xulrunner builds but doesn't do anything
            - mentioned on fisher's blog as to what it is
            - want a migration to Fx and Tb
    - Q205
        - libxul.so
            - Shrink entry points from xulrunner so that ONLY frozen entry points
              are exposed
            - Goal of having something similar to what Quicken does with IE engine
              (trident)
    - Q405
        - Moz 2.0
            - need developer support infrastructure ongoing up to this point
    - shaver working on a sort of interface between .net asseblies and xpcom.
        - says it's been tested with c# and ironpython, but it's rough
    - Narcissus
        - Want to glue everything together takes a bunch of code
        - desire to have a single object mode, reflection, inspection etc etc
        - goal to make jump from xul to C++ easier
        - lift up spidermonkey, replace component part, replace with something
          better
        - maybe at same time replace frontend with narcissus, allowing people
          to target jbm, .net, etc
    - XMLE4X
        - looks cool, I've seen similar stuff in python
        - can grab all descendants and do junk with them
        - xml objects created as necessary
        - say it'll be checked into the tree "soon"
    - canvas
        - got apple's api implemented
        - vincent hacked it out since wednesday
        - using cairo to do blending, formatting, etc
        - looking to standardize stuff through WHATWG
        - postscript-like api
        - only works on vlad's and shaver's laptop
    - xul2
        - template overhaul (!)
        - single file apps (shaver: mail it to someone!)
        - better remote-app support (e.g. being able to sort columns)
        - Drag and drop api (again, want to WHATWG w/apple)
        - looking at transparent/non-rectangular window
    - want help with wikis
        - write docs -- heh
        - say that moz 2.0 is a platform push

roc:
    - Top level window -- background:transparent gives a transparent window
    - demo of animated gif in linux
    - svg
        - shooting for enabling in 1.8 or 1.9
        - demo semitransparent overlaid butterfly
    - columns
        - hack of a sort, eventually hoping for full CSS3 compat.
    - Plan on getting approx. constant actual size for pixels
        - says it's not too hard once move to cairo is done


edit: notes are indented, add code tags so that indention is visible
Last edited by grayrest on August 7th, 2004, 6:16 pm, edited 1 time in total.
User avatar
Waldo
Posts: 596
Joined: July 29th, 2003, 8:21 am
Location: Cambridge, MA
Contact:

Post by Waldo »

See also the general overview by Chris Blizzard, which fleshes out an outline of events with some specifics.
User avatar
grayrest
Posts: 468
Joined: November 5th, 2002, 8:49 am
Location: Tribus!
Contact:

Post by grayrest »

I actually wrote this while sitting in the airport on the way home. Unfortunately, the flight was full so it took me an extra day to get home.

Dev Day wrapup
``````````````
The internals, as always are the most interesting part of what's happening. shaver's drive to 2.0 posting from a couple weeks ago pretty much sums up the foundation's direction with the core technologies we know and love.

The later discussions of the day were actually quite interesting. Vlad did a second demonstration of the canvas drawing widget. The best way to think of it is like an image that you can draw on. Once the image goes down, however, it requires a complete repaint of everything in order to change anything (in other words, lines you lay down are not addressable again like they are with svg). The canvas itself is built on cairo, is transparent, and is fully alpha blended. The implication is that it would be possible to do something like apple's dashboard on windows as soon as the canvas widget lands using the transparent window demonstrated this morning. All thedrawing code is js and they say it's a completely standard drawing api. They say they've implemented apple's functionality (except for shadowing, which is built into quartz but will be a pain for them to implement). Since cairo provides a nice gradient api, they plan on exposing that to the widget and extending apple's widget that way.

Speaking of cairo, it seems to be the hot framework for the core team. The rendering engine is going to go to cairo, which will (apparently) be much better for the engine than the current apis (gdi on win32 and whatever they're using on linux). It seems clear to me that a significant chunk of the linux desktop is coming to focus on cairo. I know that freedesktop.org (and keith packard in particular) are looking to implement X windows in cairo so that it can be easily hardware accelerated usingglitz. I'm not sure what's going on with the glitz on win32, but hopefully we'll wind up with hardware accelerated gecko on both platforms.

Nigel McFarlane's talk was fairly holistic. It's also one of the most dramatic feature requests I've ever seen. Essentially he says that it's always possible to build a better mousetrap and that we as a community shouldn't be afraid to compete (in XUL) against pre-established apps based on interface design. I agree to some extent. Since xul/css is so close to html/css it should be possible for us to tap the creative power of the web community to build beautiful interfaces. Nigel's feature requests were basically some kind of tool that would help people build xul apps, templates and trees in particular. He's opposed to the idea of having a completely graphical xul builder, but instead wants a unix-style small tools approach. I disagree with him somewhat. I believe that moz excels in cross platform RAD. I've used a few other gui toolkits, and while I'm not an expert, I think that xul is the absolute best toolkit. What mozilla lacks is a convenient way to manipulate databases and/or persist data. RDF is not a straightforward concept and, that being the only way to persist data, it presents a barrier to quickly hacking out apps. This isn't the only problem with developing apps in mozilla, but is one that would help me out. I'd like a solid set of python bindings so I could use something like SQLObject to handle this database interaction while still having the styling and flexibility of XUL for the GUI.

The other cool hack was dbaron's demo that he just checked in. The problem with userContent.css is that it applies to all websites. The new syntax, which is something like @-location-prefix("http://mozilla.org"), makes the block only apply to a specific web site. That's cool, now I can have gmail (and only gmail) in the fixed with font that is proper for mail applications.

Myk Melez is doing good work with forumzilla (which is poorly named, but...). I had forgotten how much more compelling the mail interface is for dealing with email-like objects. The most novel concept from the discussion was the ability to email a blog entry to someone else. As a person who is always directing others to interesting articles, this appeals to me. Unfortunately, not enough to get me out of firefox and into another app. Sage has problems (that are irritating me enough that I may start scratching), but is enough of an aggregator for me.

The (related) last thing I did for the day was find out why livemarks aren't an extension. I have a beef against livemarks. The entire point of RSS is to have the browser check the website feeds so that you don't have to. I tried doing this with the livemarks (setting the folder to refresh) but nothing changes when new posts come up. The need to check every livemark feed to see if they've changed combined with the fact that the subitems themselves don't show you that they're visited makes livemarks the worst aggregator implementation I've ever heard of. The one cool thing you can do with them is use a bookmarks service (like del.icio.us) to serve the bookmarks you want to share with others as an RSS feed. For this we add a new, incredibly geeky, and confusing feature to what is supposed to be a minimal browser. It's just not worth it. I've been wondering how this feature snuck in and asa told me. Apparently vlad wanted it and paid asa off by fixing about a dozen longstanding, ugly, bookmark bugs. Since nobody else understands the bookmarks code, it's a decent bargain. Is the trade worth it? Only if he keeps fixing bookmarks bugs. I mean, if the trick to getting your pet UI into firefox is fixing a dozen bugs then I'll have to get back into development again...
shaver
Posts: 5
Joined: June 14th, 2004, 9:18 pm
Location: Toronto, ON
Contact:

Post by shaver »

grayrest wrote:The need to check every livemark feed to see if they've changed combined with the fact that the subitems themselves don't show you that they're visited makes livemarks the worst aggregator implementation I've ever heard of.


Livemarks are by no means meant to replace your favourite aggregator software, whether that's an extension or another piece of software. Think of them as a way to do dynamic, remote bookmarks using a widely-deployed technology like RSS, instead of trying to push XBEL or some other undersupported thing on the world.

grayrest wrote:I've been wondering how this feature snuck in and asa told me. Apparently vlad wanted it and paid asa off by fixing about a dozen longstanding, ugly, bookmark bugs. Since nobody else understands the bookmarks code, it's a decent bargain. Is the trade worth it? Only if he keeps fixing bookmarks bugs. I mean, if the trick to getting your pet UI into firefox is fixing a dozen bugs then I'll have to get back into development again...


If Asa told you that, I think he's quite mistaken. Vlad was <i>asked</i> by Ben to implement livemarks, and did it as a way to become more familiar with the platform. In the process, he fixed a few dozen bookmarks, template builder, DOMParser, menu and favicon bugs, indeed, but that wasn't any sort of bribe -- not that it's clear why one would need to pay off Asa to get a feature into Fx 1.0! You can look at the livemarks bug history to see that it predates (even as a 1.0+ feature, IIRC) Vlad's involvement with Firefox development.

Mike
--
when the going gets tough, the tough give cryptic error messages
shaver
Posts: 5
Joined: June 14th, 2004, 9:18 pm
Location: Toronto, ON
Contact:

Re: Dev day notes

Post by shaver »

grayrest wrote: - shaver - aboout - heh


You're so cruel!

- Increased grapbics capabilities in PLATFORM not WEB


In both, actually -- we want better support (performance and quality) for Web graphics features including opacity/compositing, image scaling, SVG, etc. as well as for future platform features.

- canvas
- got apple's api implemented
- vincent hacked it out since wednesday
- using cairo to do blending, formatting, etc


It was actually Stuart (pavlov) and Vlad that got it hacked out since Wednesday, though others (bsmedberg, ndeakin) have done other patches in the past that implemented other APIs, or exposed other capabilities.

Thanks for the notes (and for attending!)

Mike
--
when the going gets tough, the tough give cryptic error messages
User avatar
grayrest
Posts: 468
Joined: November 5th, 2002, 8:49 am
Location: Tribus!
Contact:

Post by grayrest »

shaver wrote:If Asa told you that, I think he's quite mistaken. Vlad was <i>asked</i> by Ben to implement livemarks, and did it as a way to become more familiar with the platform. In the process, he fixed a few dozen bookmarks, template builder, DOMParser, menu and favicon bugs, indeed, but that wasn't any sort of bribe -- not that it's clear why one would need to pay off Asa to get a feature into Fx 1.0! You can look at the livemarks bug history to see that it predates (even as a 1.0+ feature, IIRC) Vlad's involvement with Firefox development.


It was probably me misunderstanding asa. As mentioned, I take issue with livemarks because this is exactly the type of functionality that has been pushed to extensions up to this point.

From the beginning of the bug the purpose has been for RSS integration into firefox, a job to which they aren't particularly well suited. Sage and forumzilla do a much better job as aggregators. The remote bookmarks idea is a good one and could be useful, but I don't expect many end users to set up an account for sharing bookmarks.

Apologies to vlad, I should have investigated a bit more before shoving blame around.

shaver wrote:You're so cruel!


I try not to be! Excessive oo sounds remind me that boogs are sometimes not found.
mkp123
Posts: 2
Joined: August 27th, 2010, 4:39 am

Re: Dev day notes

Post by mkp123 »

Dev day notes is interesting to read and it is an encouragement to all. The information about SQL, html/css are new to the fresher. The code given here is a good attempt to generate a new program. This type of attempt will be a way to the new inventions. Lot of new applications with Mozilla and others are discussed here. The attempt to find out new things is helpful to gain more knowledge in this field.
User avatar
LIMPET235
Moderator
Posts: 39933
Joined: October 19th, 2007, 1:53 am
Location: The South Coast of N.S.W. Oz.

Re: Dev day notes

Post by LIMPET235 »

Hi,
Why reply to a thread from August 2004.
Please don't drag up these long dead threads.
Locking.
[Ancient Amateur Astronomer.]
Win-10-H/64 bit/500G SSD/16 Gig Ram/450Watt PSU/350WattUPS/Firefox-115.0.2/T-bird-115.3.2./SnagIt-v10.0.1/MWP-7.12.125.

(Always choose the "Custom" Install.)
Locked