[ts-gen] Re: recommended C++ compiler for R package ?

[R P Herrold]

On Sat, 17 Feb 2007, mel wrote:

> I'm only at the very beginning of this work. I will look at 
> the trading-shim code (but i must confess that without 
> proper doc i feel it hard to swim in the 349 files 
> provided).

Hi, mel. I just found this email in a second tier collection 
of sorting -- please excuse my delay in replying.

Yes, we know it is daunting - we have talked through the 
architectural manual design somewhat, and needs to progress to 
the writing phase and this is on our roadmap.  The code has 
some comments, but without strong c++ skills can seem opaque.

The scripts and database .sql files are pretty well a good 
place to see all supported elements in their usage -- those 
scripts and in the Makefile the 'make test' elements represent 
a starting point to learn.  We run the 'make test', dozens if 
not hundreds of times a day, in our coding and testing 
process.

I have some personal 'cheatsheet' notes on the command verbs, 
running over 50 typeset pages so far in a PDF, and I probably 
should update, proof, polish, and push it externally as well. 
As it has stubbed out places for non-implemented elements, I 
had resisted doing such a partial release.  Perhaps I can come 
up with a way to address that; I have some ideas. ... back 
from testing -- OK -- I can comment out and regen the document 
without the unimplemented sections pretty straightforwardly. 
I'll try to turn a external release quality copy out early 
next week.

So far, I have been handling the FAQ, and the mailing list 
triage replies, escalating to Bill as I feel out of my depth.

A tutorial or introductory guide would be useful as well, so 
that a person can have a sequential and guided exploration of 
the scope and powers of the trading-shim.  We had an 
interesting discussion on the integration with R, and some of 
the possible pitfalls this week.  But doco writing is not our 
present major orientation, as we have encountered some blind 
alleys in implementing the code, and needed to back out 
approaches.

A usage guide, with more formality than the content in ./bin/ 
and ./sql/ for extending in locally interesting Underlyings, 
and such.  But presently there is not all that much material 
one has to master in those two sub directories, and it is 
commented rather well.  It has seemed to me that getting 
co-developers running and able to experiment was more 
important in my focus on the mailing list, to which I send a 
copy of this email.

Working through the INSTALL, the README and the NEWS should 
get one started and we have had those files present in each 
release from the start; NEWS is updated with each external 
release.  The regression and installation files in ./bin/ for 
getting the database are a help as well, and as I 
say, I feel are sufficiently commented.

I know it would be preferable to be done with a formal 'major 
release', and to have documentation, but then we started as a 
fast tracked developmental and explorational project, and have 
made tremendous progress so far.  The mailing list archives 
reveal pretty clearly that Bill or I will address any 
questions promptly.  Please give it a try, join the mailing 
list, and ask away when you need help.

-- Russ Herrold