On 26 June 2013 01:18, Joakim <joa...@airpost.net> wrote: > I was underwhelmed by Manu's talk: too much low-level technical detail > about the integration effort between C++ and D, not enough discussion of > the benefits of using D. The Q&A panel with Walter and Andrei should have > been an hour, or until questions petered out, and held every day of DConf. > :) >
I'd like to add a few comments here, since I guess I didn't do so at the conference. I initially really wanted my talk to be as you say/hoped. But I started down that path, and realised I didn't have anywhere near enough meat to fill an hour. The truth is, integration of D with our workflow took a lot longer than we'd hoped, and in some ways it's a miracle we persisted... We got caught up on endless stumbling blocks, but the whole time, we could see the light at the end of the tunnel. Just one last hurdle... yay... oh, no actually... just one last hurdle... [repeat]. The segment at the start of my talk wrangling compilers, runtime libraries, IDE's, debuginfo... this actually took probably 80-90% of our time. I had a lot more slides in there previously, but it felt like they were a bit of a downer, and not that interesting, so I took them out and just summarised. As a result, when I wrote the talk, there wasn't as much code written in D as I'd originally hoped, or enough time spent by the average programmer writing D code that I could draw strong conclusions of the type you'd have liked to hear about. In the end, I thought it was more valuable to discuss our hurdles, and justify some of the design points, requirements, and reasons behind language features I had pushed for. I think what I wanted as the most important take-away from my talk, was to generate discussions to make sure other companies approaching D cold, as we did, don't need to go through the same painful lead-in process in the future. Additionally, the design of our framework, which is a fairly solid piece of work, may be of interest to others looking to do similar things. It ended up with a good block in the middle, which I admit, does appear to be the focus since it was pretty hard to summarise, and dominates the slides. I hope it was interesting at least. It was something concrete that I could share. At the end of the day, you all know what D code looks like anyway. When our guys write D code, it looks more or less like you expect, and it just works. Nothing really special about it. Naturally, I couldn't discuss specifics of code written in D, since they're game-specific features, and protected by NDA. In the last segment, I did give a high-level glance over some of the things that basically all our programmers upon first contact immediately appreciated when working with D. Maybe they seem trivial, but it doesn't hurt to re-enforce that they are very strong features of the language that everyone falls in love with immediately. I guess, in summary, sorry you were underwhelmed/disappointed. To be honest, I was too, I'd hoped I could offer more. I think a lot of other people did too... but maybe next year there will be another one with an additional year's practical experience...? :)