Diff for "HowToHostASageDays"

How to Host a Sage Days

The following are just some random thoughts following one of the Sage Days conferences. Perhaps this page can serve as a place for people to jot down the lessons learned, so that future Sage Days can benefit.

• Make sure that there is someone with trac admin rights so that trac accounts can be made on the first day for all participants that don't have a trac account yet. Admin rights can be obtained by asking for them on [email protected].

• Number of talks - What should happen is something like the following. Maybe there is one big fat official talk per day, all from people who are
  1. Very famous in the field, providing perspectives like "Where does computation fit into the subject as a whole?".
  2. People who are doing active research in the field and want to start using Sage. I think the idea of a "wishlist" talk isn't quite right, more to the effect of the question asked above. After all, this is what *lists* are for...
• The official talks that do occur should have a different atmosphere than is typical for the kind of talks we usually have, e.g. in the Sage seminar at UW. It is quite disrespectful to invite someone to give a talk at a conference, and then proceed to ignore them, typing loudly and discussing things with your neighbor during the talk. At the very least, the organizer should request that people who may be doing work at the same time SIT AT THE BACK OF THE ROOM. The same goes for the small side conversations related to sprints that go on during talks...
• Official is a slight misnomer - perhaps "important" is a better way to put it, or "perspective", or "theoretical". Because every conference will have many sessions of the other type: For example (at SD7), Robert M's talk on generation, Emily's on planarity, Mike H's on development in Sage, Robert B's on 3d graphics, etc... These kinds of talks fit in well with the ones that didn't show up in the pre-arranged schedule at SD7, like Mabshoff's on valgrind, Jason's on mercurial queues, David Roe's on coercion, etc. These types of talks should be considered "optional", and follow more closely the Sage seminar format, or sprint-related talks - i.e. speakers are not at all surprised when people are hacking in the front row. Perhaps most of these kinds of talks would be improvised during the week, or planned ahead of time but not scheduled until the conference begins.
• The inevitability of all-nighters.
  1. It will happen. It will happen every night. Everyone knows this. There shouldn't be anything officially scheduled any time ever before, say, noon, because people may not go to sleep until 8:00 am.
  2. This makes the refreshments/food situation difficult, especially to the more caffeine-prone of us. For example, there was a coffee machine at IPAM, but no coffee. And when we needed coffee the most, i.e. for the final few laps of those late sprints, we couldn't get to it. I'm not sure exactly how to handle this, but it's pretty easy to accomplish, as long as it is expected in advance. Compare SD7 to SD4: at SD4, there were big pots of coffee across the hall from the lecture rooms (granted that was during the day, but think of the possibilities). This is particularly relevant to undergraduates on shoe-string budgets. Also, any kind of bulk food being on premises is better than an hour's delay to go mill about town and find food, if in fact the hours allow that at all. Coding time is precious, especially when you're on a roll.
• Trac accounts. -- If your sage days is not a sage bugs day with only die hard sage devs attending (which is likely if you are reading this page), then there is a big chance that not everybody will have a trac account yet. Since the current account creation process is quite slow (you have to e-mail to sage-trac-account mailing list and wait for someone to read that e-mail and create an account) it's good to make sure that you have someone at the sage days who has account creation right on trac. An alternative is to mention already during the registration process that people will need to make a sage trac account. If you don't do one of the previous two things you might end up having some participants that don't have a trac account the first few days of the workshop.
• "Name that trac ticket" -- this was just good times all around. Thumbs up to William for coming up with that! The organizers of the next big SD should start brainstorming a new game now ... or just start memorizing the install.log, if anyone ever wants to compete with mabshoff. By SD9 or 10, it might have to be "everyone else vs. mabshoff."
• A proposed schedule format for future conferences:
  • - First day
    • 10:00 am - Keynote lecture
    • 12:00 pm - William or someone else gives a talk on the state of Sage in general.
    • after that - sprint planning
    • throughout the afternoon/evening - optional/developer talks and sprinting
    - Each day
    • 1:00 pm - "official/important/perspective/theoretical" Talk
    • after that - sprint planning
    • throughout the afternoon/evening - optional/developer talks and sprinting
    - Last day
    • maybe no official talks this day, but wrap ups, etc.

• How to spread sage during a workshop