[RAS] acis development

[Thomas Krichel]

  Ivan Kurmanov writes

> I have not started any significant development of new features yet.

  Good news for me.

> There were some minor changes and bug fixes, and I've done them on the
> live RAS system before or soon after the migration.
> 
> Last week I've started working on integrating these changes back into
> the ACIS development. And most of that is done, just a few minor items
> left. I will soon be ready to share those patches with you, if you
> want. But they are based on a 2007 version of ACIS, and some of them
> you already have in your version.

  I want them. But I am not sure if I will be able to meaningful
  intergate them into the current version. I am very cut for time
  right now, but I can try to free a day for this. This will have
  to be this or next week. 

> The obstacles to using the latest ACIS version that are unresolved,
> that we need to clarify are:
> 
>  - does it support Storable strings in the objects table, and the
> resources table (etc.) and in the RI db?

  Yes. The change is that it uses one single serialization module
  for all. That module tries Storable as well. 

>  - are the xslt templates compatible with the RAS phrases?

  The AC templates are namespace aware. That implies that the 
  phrases have to be put into the http://acis.openlib.org/ namespace.
  That's a one-minute job. 
 
>  - does it have any other changes that might turn out incompatible
> with the current RAS setup?

  When I did the conversion of the RAS presenter to the AC presenter,
  I did not convert the citaitions screens. 

> Another thought I'm having is: do we really need the installation
> script and procedure in ACIS?  My thinking now is that it is mostly
> superflous. 

  Yes.

> We could work on ACIS code in the same files and
> directories as it is installed now, without the reshuffling that the
> install.sh does. Most of the work that bin/setup does can also be
> removed (with help of modules like FindBin, etc.).

  I still run setup, for the RI writing the configuraiton, I think
  this can't be done without the current setup, a ri restart does
  not do that (maybe it ought?, oh, no that o/s depenedent). 

> It would make developing easier. Publishing changes would be done via
> darcs repo directly, not via packaged versions.

  But packaging colud still be done when we want to give people
  a copy of the software?

  Cheers,

  Thomas Krichel                    http://openlib.org/home/krichel
                                      http://authorprofile.org/pkr1
                                               skype: thomaskrichel