Development

Last weekend wasn’t really restful this time.

Four flights, three countries, two read books, one expired visa.

Up until now all KDE developers were just abstract nicknames to me, and this is where it started getting real - at the hackathon.

The atmosphere was very inspiring and thought-provoking (including the beer ).

We discussed the present and went on for planning the future, and were doing it for two days in a row. It was a pleasure to meet such bright people and exchange ideas with them.

There is a picture of almost all participants on basysblog.

What this means for Skype? This means we will try to support Telepathy and Decibel as much as we can, and certainly keep our minds open about desktop integration this way.

Long time no see folks. 1.4 is shaping up nicely and is usable for me and Andy locally. It is also almost usable for our betatester guys, tho they still prefer 1.3.0.57 (chuckle)

Some screenshots today.

Continue reading "Updates" »

Writing about Skype is kinda pointless at the moment, because half of information is closed and other half is nonexistent.

We are doing a rewrite of codebase using Qt4.

So far we have some basic contact list stuff, with contact cards etc. And! You can switch your online status.

So it’s almost a full fledged client here, we’re about to ship it soon.

Just kidding.

I cannot say I’m satisfied with Qt4 MV model. It puts a lot of barriers, but otoh we might be using it wrongly yet :)

On the other hand, I’m rather satisfied with sigc++ signals — it works pretty well, if only missing a few nice points present in sigslot, but I can live with that.

People in closed beta - stay tuned, we’re gonna show you something Real Soon Now(tm).

Take care.

I haven’t updated the blog for a while. Autumn started here a bit ahead of time. I was helping our new developer to feel at home. And Skype development is going so fast now I don’t know where to start.

We’re heading to gold release. There are several essential problems that are holding us back. There are less than 5 of them tho, so we’re in rather good shape. Andy started his acquaintance with the code by fixing some minor nuisances, that make UI feel broken here and there, and overall disrupt comfortable work with Skype.

The non-UI side has experienced some improvement, too. We now have a better-faster buildbox and our closed beta testers receive builds for full range of supported distros at once. I plan to set up a nightly build script so that we can benefit from some more automation out there.

The global news are that we have hit 7 million users online simultaneously. And many more of them are now Linux Skype users :)

Oh, and a recent poll of Linux users have shown us three most often requested features, they are:

1. Video
2. Video
3. Video

We’re doing our best to implement it, but I think it is not going to be included in 1.3 series, so stop asking, cross fingers, and wait for 1.4 release ;)

We’ve just released X11 API demo code here (get skype-x11-testclient-20060728.tar.bz2 from that page).

One word of caution tho: the session termination code was not polished and robust in released open beta; one serious consequence of this is that if you terminate your API client application and launch it again, chances are high that X11 server will assign the same window ID to your handler and it will start receiving messages from the old session. This behaviour is fixed in latest closed beta and correct code will be released with next Open Beta update.

Please post your experience with the code here. I’m eager to hear your feedback.

With a major weather cooldown I’m also pleased to tell you that I’m finished with translations preparation — the strings are syncronized as much as possible. Quite a brainfucking task when you have +35C outside I would say.

Now I’m looking into doing a few “nice oh nice” fixes and then sketch a list of things to do for next major release (Will it be 1.4? Who knows.)

I hope to be mostly away from the PC next week, so take it naturally if I don’t respond :)

Well, this summer is damn hot here in Estonia. 31C today. No wonder you don’t hear from me a lot last week :) But here are some hot news from a hot country.

I’ve been digging i18n — a lot of strings were updated. We have around 1000 strings and many of them remain untranslated. I’m cleaning and cleaning and cleaning them up and my brain boils and then this temperature is here as well. Awful.

I collected reports from comments on this blog and from Skype forum — the feedback is pretty positive, but there are some serious problems that need to be fixed, so expect an update any time soon.

Warm laziness kicks in at this point. See you later folks, keep up the buzz.

One small note: D-BUS support is disabled by default, so that people who don’t use it and don’t needed are not affected by long startup times and so on and so on.

To bring it back in 1.2-compliant manner add the following switches to command line when starting Skype:

--enable-dbus --use-system-dbus

The reason for this is that Skype uses per-user session bus by default now, while old behavior was to use system-wide bus.

Sorry for the confusion.

…is ready. We’ll be doing some more tests and release an open beta after that. Stay tuned.

Funny thing I just got playing with improving translations is that yaml dump of original strings is just 1 byte smaller than original strings xml file, but it contains more useful information and loads a decent bit faster.

180873  skype_en.ts
180872  skype_en.yaml

(Yes, you got it, I’m doing it all with Ruby. What a joy!)

So all important things in the 1.3 queue are done now, only testing and fixing bugs left.

I’m sketching the refactor process for 1.4 — what I have to throw out, what to rewrite and what to keep. Porting to Qt4 is a major change because of MVC and cross-thread signals. The code will become much more lightweight as a result. A major speedup of chat code is planned as well, many intermediate steps will be removed.

For the 1.3 — don’t get all excited about it yet, but the open beta is close, very close.

I’ve put D-BUS 0.23 support back into latest Skype closed beta, seems to work for me.

Next step would be to put D-BUS 0.61+ support as well, since I was able to have all D-BUS support in a single Transport module it should be very easy to do.

It would be nice to add D-BUS transport support to demo Skype console as well.

I’m making a list of things to port for next release, so if you think something should be changed in Skype for Linux — say now ;)

Okay, now that I’ve ported the code, I need to downgrade my SuSE back to 0.23 somehow and it doesn’t quite look possible, given heavy dependency of HAL modules on recent D-BUS.

If I’m not back by monday this means I’ve ruined my SuSE into pieces ;)

I’ve successfully implemented X11 public API transport for Skype UI.

It is fast, reliable and in principle you can use it even across network.

I’ve also implemented a small demo application called Skype API console. It lets you interactively enter API commands, receive replies and control Skype UI. After some polishing I’m planning to release this console under open source license as an example of how X11 transport can be used with Skype API.

Continue reading "X11 success" »

Folks, we had many applications from you in the past days for the closed beta access. We got more than we can have at the moment, so please refrain from applying for some time from now on. I’ll keep you posted when we need more. Thank you all ;)

Okay, X11 IPC started working. I need to finish session support (registering and notifying clients) but that should be relatively easy now that I have message passing working.

New closed beta is out, but there doesn’t seem to be any sound improvement, which is a major PITA. But we now have a config file switch to alternate between OSS and ALSA support, so there’s a fallback at least. If you’re a closed beta tester - go download the latest beta and test it with your ALSA, I am really grateful for any and all feedback (just a simple “works/doesn’t work” will be great, post either here or in the forum).

I also want to thank all our beta testers for their patience and their invaluable help while we beat the crap out of ALSA. Thanks guys, you rock!

I asked this question in my LiveJournal for Russian-enabled readers. And I also ask this question (may be a bit ahead of time) here:

• How would you like to see the new Skype UI?
• How should it function?
• How should it look?
• What features should it provide, what should be immediately visible and what should be hidden from sight initially?
• How to make the UI “smooth”, “silk-like” and extremely handy, transparent and easy for daily use?

Post your craziest ideas about it, “spatial interfaces” and whatever else you could come up with. Take the track off the usual UI and peek into some wildest experiments with user interface design.

I really want to hear what you think.

Today was… quiet in development. Instead we had a Mac team dinner, where I was in absolute minority :)

I’ll post the photos on flickr one day (once I get the USB cable :)) There were many pics of Vi, but you won’t see them online hehehe.

Since X11 goes really bad I think I switch my focus and implement mood messages instead. Good way to think about something else (the other good way is pea cider, which shifts focus to …mmmm… alternate states of mind).

Oh, and “good” news — the latest voice engine doesn’t work at all. All attempts end with “Problem with sound device”. I don’t have any idea of what is going on yet, but I’ll give it a look today and try to figure a solution. Really don’t want to put this one out into closed beta without investigating first.

I’m revolving around couple of bugs to fix and some tests to do before we can actually do a release. I can feel compression. Events, time, life itself press against that wall - deadline, step on each other’s toes, scream and push elbows.

I really hope we fixed more than we broke during these months and upcoming release will be a relief for our users.

I’m really doing something wrong with this X11 IPC, because I cannot make it work for Linux’s sake. Gotta flush the brain and start once again.

Other than that we’re doing well. I was once told that it is a good idea to blog about something daily — what you did, what you did not, what were successes and what were failures. Daily. Every evening. Without procrastination. Maybe it will be more detailed than you would want it to make, but I think it’s worth it. It really helps to clean up the mess in my head — talking to my “cardboard fellow” or to my blog readers, doesn’t actually make much difference. I’ll try to go along this line and kick me if there’s no post tomorrow.

Screwed up with Find in chat, was fixing one thing and broke another.

The current internal chat structure as not as good as I want it to be. So in order to fix some serious crashes I need to work around existing problems in design by disabling some functionality. This time searching in chat was crashing Skype if you expanded history and tried to search while it was expanding. The workaround is easy — disable Find while history is expanding. However, with all this 100 million registered users excitement I forgot to initialize flag that enables/disables search. And now we have Find that doesn’t work until you expand history. Not Good.

Do not forget to initialize your variables! :)

And hope to see your nick along those 100M registered, or may be next 100M.

So I got kind of allowance to post a little bit about Skype development - how it goes, what problems I face, how I solve them and so on. Kinda daily boring stuff.

I’m still uncertain to what level undisclosure is allowed, so bear with me, I probably will have to edit some posts :)

First bit I’ll start off with is Public API support. Over the last year I haven’t seen much applications using Skype Public API under Linux. It was probably because of D-BUS API instability, or absence of dbus daemons running on user desktops, but anyway - really annoyed with D-BUS constant changes I decided to drop D-BUS support for next release, and while D-BUS interface will be present, it will be mostly nominal, obsolete version of dbus. Hopefully, when D-BUS is mature enough and all distros have approximately the same, compatible dbus versions installed by default — we will have it back.

Continue reading "Blogoskype" »