After the TCO, TopCoder plans to start releasing functionality updates and bug fixes into both algorithm and marathon competition software. This will be done via sofware competitions, so in addition to algo/mm systems becoming better, those of you active at software side will get a chance to win some money for helping us make them better.
Last several months were devoted to different preparations in order to make it possible (some of you could have noticed assembly contests we've run for that purpose). People who previously developed those systems are no longer working at TopCoder and it happened so that large part of knowledge about them was lost. Even putting together a code to be used for further updates presented a certain challenge, which now is close to be solved.
At this point we think we have a codebase which resembles the current competition systems very closely (we know about a couple of minor differences, but almost everything is the same). While we've done quite a lot of testing, we still feel it would be useful to have even more extensive testing.
Therefore we would like to deploy this new code temporarily instead of the current systems and use it to conduct a test SRM. The more people compete in this, the more feedback we will be able to gather, so we invite and kindly ask all of you who can take part to do this tomorrow.
Some more details:
- We will reuse problems from some old SRMs (between 200 and 400). Exact problems to use will be chosen more-less randomly.
- The exact time of the SRM is not yet known, since we exactly don't know how much time it takes to do the deploy and some difficulties can potentially arise during this process. In fact, it's even possible that we won't be able to conduct this SRM tomorrow due to some unforeseen difficulties, though of course we'll do our best to conduct it. We will announce the exact start time in this thread at least 2 hours before the actual start. It is almost certain that start time is 2 PM or later.
- The SRM is not rated. We can't guarantee that SRM runs smoothly, since the code has not been tested very extensively and in fact, it is the purpose of this SRM to find out problems (hopefully and most likely, minor ones) with this new code.
If you have any questions, please feel free to ask here.
Monday, September 24, 2012
Publicado por vexorian en 4:55 PM