more books || MAIN || run don't walk

January 27, 2005

ask asa and he'll (eventually) respond

I've finally made a minute (140 of them, actually :-) to put down some answers to the latest installment of Ask Asa. I'm not necessarily the best person to answer all of these questions and I don't have time to consult all of the authorities but where I could, I asked around for some input from others. Maybe in the not too distant future we can revive the Developer Chats that Chris Nelson launched on #mozillazine way back when. Until then, I'll do my best.

OK, on to the Q & A:

backpack asked, "I was wondering what sort of additions were planned for gecko. For example, is there any timeframe for getting SVG 1.0 'finished' and into the default firefox builds? Also, are there plans to implement CSS 3 whenever the spec is final? That sort of thing."

Well, backpack, you've asked just the kind of question that covers an area I'm clearly not an expert. From my role on drivers@mozilla.org I've been involved in some of the planning and scheduling discussions around SVG and it sounds to me like a plan has emerged for what of SVG will happen and the beginnings of a plan for when. SVG depends on quite a bit of rendering (GFX) work too, so it's not just the simple question of "when will we have SVG"? Robert O'Callahan has posted a roadmap to the Mozilla wiki that outlines a plan for updating our GFX layer with Cairo, which would provide a substrate for the cross-platform Mozilla SVG implementation. As to when it will happen, I believe that the Cairo work that will give us a snazzy new graphics layer, including all of those cool zooming and blending features, is on the Mozilla (the platform) 2.0 list so I'd hope to see much of this work happening in the 1.9 timeframe. With CSS3, we've been implementing pieces of CSS 3 for ages. As you mentioned in your question, the spec is still a long way from complete. Given its incomplete state, dbaron offered me the estimation of 40%, we have been implementing it piecemeal and with a focus on areas where our app needs the feature. There are a few exceptions and we do have some CSS 3 support that Web developers today can use like some of the pseudo-class selectors, opacity, and columns.

A`ja asked, "Anything you can tell us about plans for addition of X+V 1.2 and CSS3 speech capabilities to Moz browsers, ala Opera 8 beta?"

A`ja, as far as I know, there aren't any plans to add speech capabilities to the Mozilla apps.

XeroCool said, "How will the future of FF look like?"

Well, that's a huge question. Let me try to focus it down to something more near-term. We've got a Firefox 1.1 coming up this spring/summer and the goals for that release are to move to a contemporary Gecko core (the rendering engine that shipped in Firefox 1.0 is now approximately 7 months old and will be nearly a year old when 1.1 ships.) This update to the 1.8 Gecko will provide new layout features, improved website compatibility, better performance, and hopefully stability. In addition to the Gecko improvements, we'll probably see some big improvements to the Options/Preferences window, both backend and frontend, as well as a heap of Macintosh bugfixes that should bring it much closer to the Windows and Linux builds in terms of usability and overall quality. After 1.1, we'll be looking toward the 1.5 release which is just now being planned. Look for major innovations around our new unified storage code and updated graphics capabilities. The future of FF looks good :-)

dolphining had a couple of questions. The first was "You mentioned Web Forms 2.0 in the other post. What are the plans for it, e.g. how soon (after it's finished) can we expect it, is anyone working on it already? And if you know offhand (though I should probably ask Hixie...), how soon will Opera and Safari support it?

Another question where I don't have great answers. Web Forms 2.0 is still in the defining and planning stages, I think. It's on the Mozilla (the platform) 1.0 list so I'd expect work to happen on it this year. I don't believe you'll have to wait for it to be finished before you can play with it -- unless someone has plans to change the Mozilla incremental development process. I don't have any information about Opera and Safari's plans on anything.

dolphining's second question was, "What bug currently, for whatever reason, irks you the most?"

Talk about a difficult question :-) I've been reading bugs since 1998 and for a number of years, I read almost every bug that came into the system. I'm irked by literally hundreds of bugs and couldn't possibly nail it down to one. I'd be lame for using that excuse not to answer your question so I'll mention a few that aren't necessarily the worst, but that popped into my head immediately. I really hate that we don't have a mechanism for setting default browser on Mac. I'd love to be able to delete attachments from mail messages. I'd like to see the network error dialogs replaced with an error page. Hrm. Wait. Those are actually feature requests ;-) There really aren't any bugs in Firefox or Thunderbird that really irk me. I do have a couple of bugs in Bugzilla that irk me but I'll bet no one's interested in those. If you are, feel free to look up the bugs I've reported or am cc'd on in that Product.

Ludovic Hirlimann says, "You recently said Mac builds tooks hours to complete, could you give use the specs of the machine doing those builds ?"

Well, I was probably exaggerating a tad. We do have some flashy new dual G5 Macs that QA and development inside the MoFo office are using and those are considerably faster. The older ones can be quite slow. Imola, for example, makes a Firefox branch and trunk builds completes in about an hour. It's a quicksilver 867 MHz G4 with 2 MB cache and 512 MB RAM. But completing the build isn't the only problem. We also have to upload the build and the talkback symbols and that can take up to an hour so if we're sitting around waiting on the next build to come out with the latest checkins, it can take two or more hours for Mac.

My lovely wife, Deanna asked, "What...is your name? What...is your quest? What...is your favorite color? What...is the capital of Assyria? What...is the airspeed velocity of an unladen swallow?"

Hrm. What do you mean, an African or a European swallow? ;-)

minghong asked, "When will Mozilla support ruby characters? :-P"

That would probably be bug 33339. I don't think anyone has an answer to that. Perhaps someone who cares enough will come along and make it happen.

Martijn had quite a few questions. Let's start with the first and work our way through :-) "I'm seeing a lot of work done on SVG and XForms. Who are those guys working on that (they are specifically working only on that)? And why are they doing it (now)?"

The guys working on SVG (and it's dependencies) include Tim Rowley, Robert O'Callahan, Jonathan Watt, and probably a few others (certainly they're getting some help from our core layout folks like dbaron, bz, and others.) I haven't been following this closely so anyone that has, feel free to chime in with more info. Bryan Ryner and Darin Fisher were the two I know of working on XForms. More recently, Doron Rosenberg has been helping too. I'm guessing that since Darin's gone to Google, that will just be bryner and doron now :| though I'm not sure of that. Some of these contributors are paid to work on it, others aren't. I don't think any of them have SVG as their exclusive contribution to Mozilla.

"Can I help in those areas, with testcases and stuff? (if I find the time)"

I'm sure you could. They have a project page at http://www.mozilla.org/projects/svg/ and lots of bugs in Bugzilla. You can probably connect up with those folks through one of those avenues. For XForms, you should probably email bryner or doron and see if there's anything they need.

"Where can I download those builds with SVG/XTF/XForms enable? (Did I forget something?)"

You can occasionally find builds with SVG posted at Mozilla's FTP site. I think the links are available at the SVG Builds page. I don't believe there are any XForms builds available yet. Doron says that XPIs are coming soon though.

"I've submitted a few simple patches for Firefox code, but I've noticed that reviews are rather slow. Is that normal?"

Yes. That's normal. There are currently about 1,100 patches in Bugzilla awaiting review or super-review.

"Also, the people who do reviews for Firefox is not listed somewhere, like for Mozilla. So, it's not really obvious for me who should do reviews/who is doing reviews. Wouldn't such a review list be handy?"

They certainly are listed. You can find reviewers at the Module Owners page. An owner or Peer is able to do reviews. For Firefox, those modules would be Toolkit and Browser.

Ivan Icin says, "You publish stats frequently. I was wandering, are those percents applied to total number of Internet users, or to number of page hits?"

Actually, Ivan, it depends on which stats I'm publishing. The download stats are just that, completed Firefox and Thunderbird downloads. The stats from other organizations like WebSideStory and OneStat are usually based on website tracking.

Josť Jeria asks, "What happend to devedge.netscape.com? They shut it down, and then it was back up again? And now gone again? I think many web developers would be very happy to see it back again.

Josť, AOL, who owned all that content pulled it down. As soon as that happened, the Mozilla Foundation looked into it and Mitchell Baker started work to try to secure that documentation. I think she's got a post coming up on this but things are looking really good.

jens.b asked a couple of questions in his comment. The first was, "On some developer day slides, it was mentioned that Mozilla would move to a graphics engine library called cairo, and that a new canvas implementation was started as part of that work (i.e. not based on previous work for bug 102285). However, I miss some information about this - progress, who's working on it, target gecko milestone etc. Do you have any info on that? I wonder why, unlike other stuff planned for Mozilla-the-platform 2.0, this new canvas implementation doesn't seem to have any up-to-date details in the mozilla wiki, the corresponding bug, newsgroups..."

I believe that this implementation would wait on Cairo support in Mozilla and would hope that the experts, Stuart and Vlad, would pick that work up when the substrate is primed. I don't have much more to offer here, sorry.

The second question was "I often wonder how users, interested developers or soon-to-become-developers are supposed to know whether a bug is intended to be fixed (imagine a "yes, we want this, feel free to work on it" tag). For example, many bugs in Firefox are just assigned to Ben Goodger, but with no indication he or his peers did ever look at it. I mean, we have ASSIGNED and RESOLVED WONTFIX, but these bugs are in between those, in a state that could be described as "UNNOTICED" or "UNDECIDED". Given that bugmail is not always read, how can one bring them out of this state?"

This is a good question. We don't have a great system for making this clear in Bugzilla. The first thing I'd suggest is to make contact with other active developers in that area and find out where they're looking for help. We've always had more bug reports coming in to Bugzilla than our developers can possibly stay on top of if they're going to get any work done. For this reason, we try to harness the community to clean up and triage bug reports, to nominate major problems, and certainly to contribute patches for problems they can fix. Utilizing the community of people who care about particular issues is critical to our successes. But, yes, you're right. It's not often easy to tell whether or not the issue is one that we (those responsible for the product decisions) want the changes. If you're unsure and you can't get info from the module owner or peers, then I'd move on to a bug that's clearly in need of fixing. We have enough of those that no one who wants to fix bugs should feel like there isn't a good place to start. Avoiding "features" and devisive changes is probably a good idea since those are most likely to cost you a lot of time without a lot of benefit.

Josť Jeria says "How should bug reports that are filed with 1.0 be treated? For example typical "firefox crashes on this website" kind of bug. If it works on the trunk with the same system, should it be marked as WFM right away? Chances are that the bug was fixed long time ago."

We're not going to be releasing any 1.0.x releases with fixes that would break Gecko compatibility with 1.0 or Mozilla 1.7.5 -- if at all possible. If the bug is working on the trunk and it's not security related, then go ahead and resolve it as Worksforme, unless you can locate the actual bug where the fixed happened, in which case you might resolve it as a Duplicate of that.

Gids says, "Similar to Josť's question: I'm looking to help out more with Bugzilla, when I check the 'bugs already reported today' I often see bugs like 'foo.com doesn't look right in Firefox' filed for example under Firefox General. Most of the time these pages are far away from any W3C spec, so can't be expected to render as in IE. What should happen to these bugs should they, for example, be marked INVALID or moved off to Tech Evangelism?"

Gids, this is a bit more complicated that it would at first seem. To correctly process them I'd suggest a few steps. First, the easy one. Check to see if it's a dupe of an existing layout or evangelism report. My latest calculations show that more than a third of bugs reported these days turn out to be duplicates. If you don't find a dupe, then the next step is to come up with a simplified testcase and look for any layout or evangelism bugs that share the same problem. Once that work is done, comes the hard part, determining if the issue is something we should try to fix or not. For that you often have to consult one of the experts. If you've got the simplified testcase, though, it's usually not too difficult to find a layout person to look at it. I wouldn't mark them invalid, though. They're either a layout issue or an evangelism issue and until we get the "report a broken website" tool hooked up (any day now) we're still accepting evangelism bugs. Our process for accepting layout and evangelism bugs will probably change, though, once we have our new tool

David Naylor asks "When and why was the idea of including a couple of extensions with the FF download abandoned?"

It wasn't. We ship the DOM Inspector as an included extension and we may include others if we find ones of sufficient value and quality.

Neil Paris wonders, "Any chance of getting the graph from December (http://weblogs.mozillazine.org/asa/archives/007184.html) updated (or a similar one for January, maybe?) I think it would be interesting to see."

Yeah, I'm too late at responding to the Ask Asa questions. I've already posted some updated graphs. You can find them here and here.

perko says "Was there a conscious decision to not add a download address bar in the download manager? Is one in the works for Firefox 1.1?"

Yes, all of the UI decisions about Firefox 1.1 were conscious :) I don't believe that this is planned for 1.1.

pipe asked "How many of the Secunia vunerabilities would you expect to be fixed for Firefox 1.1?"

We've already fixed one or two of them already (at least the window injection one) and we'll fix the ones that really matter and fix them in time for 1.1. There are a couple that really aren't that important, though, and I'm not sure about those.

larfnarf asked "could you please explain the way that the Mozilla FTP/Firefox distribution works? I was under the impression that the org went through a lot of work to get mirrors for firefox... Also, I was under the impression that in the early days of the release that offical counts of downloads took some tabulating.. Is there now a new system that makes counting downloads easier?"

The wonderful folks at OSU's OSL have been working on a tool called Bouncer which is, to put it real simply, a mirror management tool. We serve our releases through a fairly large network of mirrors (somewhat above 50) and this Bouncer tool has some great features including the ability to easily add and remove or disable mirrors, automatic scaling and routing to the most appropriate mirrors, hash checking to insure file integrity, etc. I has mad counting downloads considerably easier and taken some of the guesswork and estimation out of it but the tool is really more about managing traffic and load than providing Asa with lots of good stats. When it's a little bit further along, I'll ask for more in terms of reporting.

squegole says "I was wondering if you could elaborate on an recent post of yours regarding the need for rich text editing and other textarea field enhancements. Have you shared this idea with others in MoFo circles? What would it take to get such features incorporated into Firefox? For example, is it realistic to hope some features of this type might be integrated before, say, version 2? Whilst on the topic of future technologies and Firefox, is integration of XForms or Web Forms 2 confirmed?"

Yes, squegole, I'm sure that the need for better rich text editing and other text area enhancements is already well known to others at MoFo. It's been an area of discussion for years. My post was really trying to approach the issue from a "what would make blogging better" standpoint but the idea of better text editing is not new. I think that it's possible we could see something like Contente Editable in Firefox 2.0. Some of the other suggestions, I'm not sure. That's one reason I was soliciting input here at the blog. I'd like to find out how much interest there is and if anyone's interested in helping to make some of those features happen. In response to your second question, yes, integration (in some form) of XForms and Web Forms 2.0 are planned. Whether they'll be in upcoming Firefox releases, I don't think that's been decided yet.

OK, that's it for this round of Ask Asa. Hope you all found it interesting. I'll field follow-ups for this issue in the comments and hopefully post another Ask Asa sometime in February. Posted by asa at January 27, 2005 09:12 PM

Comments

Asa, when people have minimal testcases and it's been confirmed that a page doesn't have a layout problem, they may want to check against the HTML parser bug list for duplicates, as our imitation of IE parser quirks is not quite perfect.

Posted by: Blake Kaplan on January 27, 2005 09:44 PM

Asa,

You mentioned that you "[...] do have a couple of bugs in Bugzilla that irk [...]" you, but indicated that you don't think anybody is interesed in those. That's not completely true. While I could search for those that you reported or are cc'd on, it may be helpful if you (or possibily the Mozilla Foundation staff) put together a short list of things you'd like fixed/changed with more priority. Especially being that you have two webtools developers on staff :).

Posted by: Jake on January 28, 2005 12:40 AM

Thanks Asa!

Posted by: Martijn on January 28, 2005 01:13 AM

Jake, yeah. But very few people use Bugzilla like I do and it's simply not worth catering to my needs given how niche they are ;-) Bugzilla is awesome and you all have done an amazing job taking this once humble project to a whole new level.

--Asa

Posted by: Asa Dotzler on January 28, 2005 01:16 AM

"I'd like to see the network error dialogs replaced with an error page."

If the mighty Asa cannot make this happen, what hope is there for the rest of us! I tried to search Bugzilla but it did not work out for me, if you are able to link to the relevant bug for this i would be very interested to read the inside story. Cheers.

Posted by: db on January 28, 2005 03:36 AM

"I'd like to see the network error dialogs replaced with an error page."

In about:config, set browser.xul.error_pages.enabled to true. There is an annoying bug involving the interaction between error pages in history, which is probably the main reason it isn't enabled by default in Firefox. Bug 216466 covers enabling it by default in Firefox and depends on several other bugs.

Posted by: Jesse Ruderman on January 28, 2005 03:53 AM

XForms stuff is led by Aaron Reed (he's leading the tech there, according to mkaply). Other folks I have seen on that subject are Allan Beaufour, as well as at least one from Novell in addition to those mentioned by Asa. XForms is really a tough beast, with extensions to XPath, dependency graphs there, too, forms stuff, schema validation and what not. Not counting reading and extracting some sense out of the spec itself.
The Mozilla XSLT team (peterv, sicking and me) is trying to help out in XPath land, too. At least as far as integration and separation of the XForms code in/from the XPath code goes.

Posted by: Axel Hecht on January 28, 2005 03:59 AM

db: error pages are implemented and you can turn them on with a hidden pref, but they don't work well enough to be turned on by default. The meta bug for error pages is bug 28586, and if you look at the dependency list you can see the stuff that still needs work to get them to be good enough.

Asa: "You can find reviewers at the Module Owners page. An owner or Peer is able to do reviews"

You only answered half the question though - the useful thing to know would be which of those people are actually doing reviews. There must be a better way of managing reviews than stacking up loads of review requests against people that won't look at them, and then having other people occasionally shuffling reviews around. Of course, having a better system for the review queue would only help if people are actually doing reviews.

Anyway, thanks for the answers :)

Posted by: michaell on January 28, 2005 04:02 AM

Hey Asa. If you do want to let us know a few of those bugs that you reported or are CC'ed on that you would *especially* like fixed, we have a lot more manpower for Bugzilla these days...

Since you're one of (if not the) most frequent user of b.m.o, I think that:

(a) You deserve a few changes
and
(b) You represent the "ultimate" user (because of the frequency with which you use the tool), who sees every little flaw that any Bugzilla user might experience and dislike. So fixing things that particularly annoy you probably also would improve the experience for a large number of users.

-Max

Posted by: Max Kanat-Alexander on January 28, 2005 04:07 AM

Asa, thanks for taking your time to answer, but actually it didn't help :).

My qusetion was intended on the stats of OneStat and WebSideStory, but unfortunatly, the fact that stats "are usually based on website tracking" doesn't resolve the issue. They set cookie (at least OneStat), so they can state percents of Internet users based on that (maybe and IP address too) and user agent of browser, and again they can state percentage of page hits, based just on user agent information they collect.

These numbers should hugely vary, so it would be nice if you could cleat this issue here or in your next stat post.

Posted by: Ivan Icin on January 28, 2005 05:25 AM

Ivan: You can probably find the answer to that as easily as Asa can - he links to the stuff he's talking about, and I don't think he has any in-depth detail on the stats beyond what they've published. WebSideStory's stats are percentages of usage (i.e. page hits) rather than users. I'm not sure about OneStat, but I would guess that they are also usage - trying to figure out a count of human beings is pretty hard.

Using cookies and IP addresses might help to some extent, but use of proxies, cookie settings, people using multiple computers or multiple browsers, and a bunch of other factors make it pretty much impossible to come up with useful user figures for even one site, let alone across the wider web.

Posted by: michaell on January 28, 2005 05:44 AM

Asa is a cool name. Who named you? What nationality are you? Did you know that there are many old Mennonites with the name Asa. There is also Asa Hutchinson from the Homeland Security Dept., Asa Arronson from a NY TV station, and the character Asa Buchanon on a soap opera. Do you know of any other Asas? I know one other, he is 23. I am 9 years old (and I ROCK!!!)

Posted by: Asa Cohen on January 29, 2005 10:56 AM

Hi, Asa Cohen.

It's good to hear from another Asa. We are uncommon as I have not met,
in person, another Asa yet. I have talked with a couple of Asas via
email or on the telephone but have not been face to face with one yet.


Yes, I know of the 'famous' Asas that you mention. I also met a woman
from Sweden named Asa (except the A character was actually the "a ring"
character with a little circle above the A and her name was pronounced Oh-suh).

My mother just liked the name Asa so that's how I got it. It is not a
traditional family name, and my parents didn't pick it because of its
biblical origin (Asa, King of Judah -- also means "healer" or
"physician" I think) I did know there were Asas in the Mennonite
communities. I grew up not to far from a Mennonite community.

How did you come to have the name Asa? Do you put the accent on the
first syllable or the second. Sometimes I am "A-suh" and other times I
am "Ace-uh". Which do you prefer?

Nice to meet you, Asa.

--Asa

Posted by: Asa Dotzler on January 29, 2005 11:50 AM

Asa, do you know when localised nightlies will be available for the trunk?

Posted by: Greg K Nicholson on January 31, 2005 12:14 PM

Post a comment