Set in the royal surroundings of Edinburgh, this year's PLDI was my most productive conference in terms of networking with people. Being quite close to graduation, this was especially helpful.
One of the most interesting and well delivered talks for me was Martin Rinard's recovery shepherding work. The whole idea of continuing to go on executing even when things go wrong is hard to conceive in practice, but Rinard's excellent oratorship could have swayed some members in the audience :) I had nice discussions with Vikram Adwe, Shriram Krishnamurthi, Yannis Smaragdakis, and got some feedback on my work. I also ended up meeting some local student volunteers for dinner after the SRC session, and got a flavor of Edinburgh student life style. The SRC session, though long and tiring, ended up being quite useful, generating a bit of a crowd and feedback.
The idea of having a one minute advertisement for upcoming talks worked out very well. Conferences with parallel tracks should try and adapt this. Previous conferences have tried poster sessions etc., to achieve a similar effect, but according to me, the ad-style works out very well.
Few of the talks were very refreshing with their ideas. The one on differentiating between shape-types and material-types in Java, and the large scale study that developers always separate between them was a pretty cool observation. Another one showed that testing for compiler bugs by eliminating input dependencies yields impressive results. MSR's tree-DSL for specifying and manipulation tree-based languages called Fast was quite interesting, and had a nice web front end to play with.
Overall, PLDI was a very fruitful experience, and I am looking forward to the next one at Portland!
One of the most interesting and well delivered talks for me was Martin Rinard's recovery shepherding work. The whole idea of continuing to go on executing even when things go wrong is hard to conceive in practice, but Rinard's excellent oratorship could have swayed some members in the audience :) I had nice discussions with Vikram Adwe, Shriram Krishnamurthi, Yannis Smaragdakis, and got some feedback on my work. I also ended up meeting some local student volunteers for dinner after the SRC session, and got a flavor of Edinburgh student life style. The SRC session, though long and tiring, ended up being quite useful, generating a bit of a crowd and feedback.
The idea of having a one minute advertisement for upcoming talks worked out very well. Conferences with parallel tracks should try and adapt this. Previous conferences have tried poster sessions etc., to achieve a similar effect, but according to me, the ad-style works out very well.
Few of the talks were very refreshing with their ideas. The one on differentiating between shape-types and material-types in Java, and the large scale study that developers always separate between them was a pretty cool observation. Another one showed that testing for compiler bugs by eliminating input dependencies yields impressive results. MSR's tree-DSL for specifying and manipulation tree-based languages called Fast was quite interesting, and had a nice web front end to play with.
Overall, PLDI was a very fruitful experience, and I am looking forward to the next one at Portland!
No comments:
Post a Comment