-- Start: Sunday, July 24, 1994 12:10:02 pm DUtime (EDT)

DanielP turns Recorder on.
BillH says:
Do we have an agenda to discuss?
DanielP says:
not really
PeterMR says:
Does anyone still meet students in the MOO?
I have missed my last two but expect to be in on Tuesday
BillH says:
I still meet students in the MOO
PeterMR says:
any impressions?
DanielP says:
Like I said a long time ago, these office hours are turning to be just like rl offices (few students and always the same onw)
BillH says:
Most of the time no students show up.
PeterMR says:
There have been one or two postings to the discussion which show there is still some activity
BillH says:
Are people meeting on the MOO to discuss project work?
PeterMR says:
I have no evidence that they are...
laurap says:
Jim once told me that it was too hard to arrange a meeting time for the String project (different timezones, different schedules).
(DanielP thinks: we would need a few more tools to make effective use of this env for collaborative software work )
laurap agrees with Daniel
PeterMR says:
I think we especially need to be able to share code and other text resources
DanielP nods
PeterMR says:
I like Laura's idea of a shared project management system *very* much
DanielP says:
is someone using w3_emacs?
laurap [to PeterMR]:
Thanks -- I hope to begin working on that.
laurap [to DanielP]:
I could start it up, but it takes a minute...
rodrigc says:
Just out of curiousity, is there a platform(s) that most of the students are doing their work under?
DanielP [to laurap]:
I would like to have more info about your project management system
laurap says:
It doesn't actually exist yet -- it's the proposal I sent out last week. The idea is that there will be forms that people can use to lock project files so they can edit them, then check them in with comments, etc.
At this point the most technically difficult point is finding a way for people to upload files conveniently -- I just (Friday) wrote a very primitive w3 client that you can use to "post" a file to a form, but it only works on unix systems.
PeterMR says:
if we don't have a specific agenda I'd like to chat about how we 'keep going' over the summer
laurap says:
other alternatives are for people to type in (or cut & paste in) files to a form, or to ftp it somewhere.
PeterMR says:
but after laurap has finished...
laurap says:
there's also the issue of authentication -- I believe they're going to start making provisions for password-based authentication on uu-gna.mit.edu, but I don't know how passwords are distributed.
However I'm planning on doing something to record each change (perhaps send mail to the project mailing list), which I hope will make authentication less of an issue. And of course previous versions of everything will be kept (with RCS).
laurap is finished :-)
PeterMR says:
one problem with WWW is downloading a specific set of pages. Maybe the form has to allow users to specify which pages they want to download and then make tar.Z
laurap [to PeterMR]:
good idea -- make that file dynamically when people want to download it, rather than when the files change.
PeterMR says:
right - and provide the users with diff/merge stuff that they (e.g. me!) find difficult
laurap [to PeterMR]:
my original idea was to have a giant tar.Z file with everything including the old rcs versions of everything.
DanielP has disconnected.
laurap [to PeterMR]:
you mean the equivalent of rcsdiff?
PeterMR says:
I don't mind giant tar.Z at all! but we have to make a new one after every change
also if there is any hierarchy in the structure it helps to have tar.Z at intermediate levels
laurap [to PeterMR]:
yes that's true -- also it just occurred to me that there may be people who don't have tar...
PeterMR says:
if they don't have tar (e.g. on Borland) they have to use gzip? It's universal I think
well at least on PC and Mac
well at least on PC and UNIX
rodrigc [to PeterMR]:
I think you can get versions of tar and gzip for DOS....I dunno about Mac
laurap [to PeterMR]:
yes, that should work, if they have an equivalent on the Mac (in general, GNU software doesn't run on the Mac.
BillH says:
The Mac has tar and gzip programs
PeterMR says:
the cat has just caught a frog... back soon
laurap says:
okay, then tar and zip should be enough.
rodrigc says:
What does tar let you do?
laurap [to rodrigc]:
Collect a bunch of files together. People would use tar or zip for the same reason.
Assistants of the local psychology institute arrive to cart DanielP off to their dream-research labs.
(BillH thinks: tar allows you to archive files into one file )
rodrigc [to laurap]:
What would be the advantage of using tar over say something like PKzip?
(BillH thinks: zip compresses; I don't think tar compresses )
PeterMR says:
after tar you can compress with compress -> *.Z or gzip -> *.gz
laurap says:
Unix systems come with tar.
laurap [to PeterMR]:
who won? :-)
PeterMR says:
we did! it's back in the pond
laurap [to PeterMR]:
Congratulations!
PeterMR says:
I suspect that if we had a class library, then there might be a tar.Z for each 'family' of classes - you wouldn't want to download *everything* after one change, but it's safest to unload the whole of a class where some change has been made
laurap [to PeterMR]:
Perhaps you should have a tar.Z file corresponding to each .a file...
PeterMR says:
yes, I think that is a good idea. I used to p;ut everything in ONE *.a but it is up to several megabytes now
I think we have just about enough critical mass to keep the project idea moving gently after the end of the formal classwork
laurap says:
do you mean the projects in general or the String project in particular?
PeterMR says:
I meant in general. I think that perhaps the String project can be used to catalyse the 'rules' for project work and that we should then start gathering in more ppl from the newsgroups if they want to work on other classes. After all we have a *lot* to
...offer and can now require a gentle contract from new partcicipants
laurap says:
I agree. I'd like to get the collaborative tools worked out on the String project people first before inflicting them on new people.
(laurap thinks: "them" refers to the tools, not the String project people :-) )
PeterMR says:
We shouldn't be too ambitious. After all , no one else out there has got a collaborative C++ project going and we are on reasonably firm ground with String. We can keep adding extra member functions easily without requiring great re-edits
I think it's perfectly reasonable to say that we have worked out how to *start* a project with reasonable chance of success.
laurap says:
yes -- I can't imagine anything new that you'd want to add that would require a change to the underlying structure of the String class.
PeterMR says:
the sorts of functions are things like 'substituteTabsBySpaces' ... these are enough of a challenge to write but don't cause serious knock-on elsewhere
laurap [to PeterMR]:
Were you about to contrast starting a project with other project-related issues?
PeterMR says:
The challenge comes when we have ca 500 member functions for String! (which is not impossible). We have to document them nicely
I was really thinking that the greatest on-line discussion is now about the projects and we should keep that going
have you looked at the matrix class project? Discussion is quite lively - they are starting from first principles
e.g. wondering how to invert a mtrix
laurap says:
No, I haven't -- I'll check it out later.
(laurap thinks: or how to tell whether a matrix is invertible, maybe. )
PeterMR says:
it has to be square, and it has to be non-singular
Coming back to the coursework, how many 'weeks' do we have left?
laurap says:
I *think* the most recent proposal was to end it August 15.
PeterMR says:
I haven't really kept track, but we are on course to do polymorphism and then templates?
laurap reads a notice on Bulletin Board.
rodrigc [to PeterMR]:
"The matrix project is pretty quiet. Only about 2 people are actively involved in it.
PeterMR says:
ah. but 2 is a reasonable number since others might be reading it
(laurap thinks: that's one more than the CGI project. )
PeterMR wonders whether a number of students have just read the book to the end and then effectively 'left'
laurap says:
I think we have the current chapter on inheritance, then one on virtual functions, then there's the template chapter and chapter 10 (whatever that is) and maybe some material by Grisha about exceptions.
PeterMR says:
templates and exceptions are tough as they are machine-dependent. (probably). It woudl be nice to get some feedback
laurap [to PeterMR]:
You may be right, especially since the pace of the course was pretty slow at the beginning. And we get mail from people not officially in the course who read the book.
Yes -- I'm not sure how much of exceptions the new g++ has implemented. (Marcus says the new g++'s handling of templates is much improved).
PeterMR says:
... about what compilers work. I think it's critical that for 'next time' we are able to say to students: please compile this example and report your experience. I think that we managed that a little at the beginning where we had quite a lot of f/back
I will have to download the new g++ (mine's quite old). But we have to be careful not to be g++-specific
until it runs on every machine in the world :-)
laurap says:
I agree! I think that templates and exceptions will need lots of testing/notes.
PeterMR says:
I think these chapters would be best by stimulated discussion. Perhaps we could post some examples and ask the students to test for compilation. BTW have any of you *used* exceptions?
laurap says:
I've been afraid to :-). The documentation for the g++ I'm using says they're not fully implemented yet.
PeterMR says:
stimulated discussion ==> we actively post to the discussion group
rodrigc [to laurap]:
"Do you know anything about the stability of the new version of gcc? Version 2.6.0?
laurap [to rodrigc]:
When it was announced, they said there would probably be a 2.6.1 within a couple of weeks.
PeterMR says:
I also think that exceptions require you to think in a somewhat different way. I wouldn't be too happy about publicly saying how they should be used, but I'd be quite happy to theorise!!
laurap says:
they do seem like a really useful concept.
rodrigc has disconnected.
PeterMR says:
look
I think some of us and the students will be 'going on vacation' fairly shortly so we have to make sure that the course doesn't just 'fade away'. I think we should try to make sure that on Aug 15 or whenever that we have a reasonable feeling of achieveme
...achievement and ideas for the future
I shall need to go shortly. Would one of you turn the recorder off and mail marcus. BTW I *might* see marcus again next w/e - he comes to London often
Assistants of the local psychology institute arrive to cart rodrigc off to their dream-research labs.
laurap says:
Okay. I agree about not wanting the course to just "fade away". Any ideas on how to achieve a sense of completion?
BillH says:
First, we could tell the students the date the course will end.
laurap [to BillH]:
Yes, that would help. :-)
PeterMR says:
I think Tina's report will be important. Also I think we should describe this course in C++-FAQ
I think that we should try to get as many '1-page' replies from students as possible
Then we should list the ppl involved in the projects and their continuing achievements.
laurap [to PeterMR]:
1-page replies to what? Tina's survey or something else?
PeterMR says:
Then I think we could also hand out 'awards' to students who have made particular contributions: most useful technical contrib, bravest student, most useful reference item, etc
Just a bit of fun, but a sort of 'roll-of-honour'
laurap likes Peter's suggestions.
PeterMR says:
after all, *participation* is worth something and I think thatr GNA would approve. It sets the scene
for example, I had already said that I would write to Diego's supervisor saying how much he has helped. he said he would appreciate it., and will let me know when
also there are some students who , because they came to the MOO regularly, have helped keep our spirits up
In fact I suspect the list would come to about 15 who had made significant and regular contributions of some sort
Anyway I have to go now! See you around...
laurap waves Peter
BillH [to PeterMR]:
Bye
PeterMR has disconnected.
BillH [to laurap]:
Do you have anything more that you want to discuss?
laurap says:
No, I think we've pretty much covered everything. Do you?
BillH says:
yes
laurap listens expectantly.
BillH says:
Yes, I think we covered everything.
laurap says:
Oh, I misunderstood. I guess we can adjourn then.

-- End: Sunday, July 24, 1994 1:19:56 pm DUtime (EDT)