source code status
rd at sanger.ac.uk
rd at sanger.ac.uk
Mon Jan 30 10:52:01 EST 1995
A new intermediate code version is available at ncbi, in the file
/repository/acedb/test.source.3.6a.tar.gz. It contains a number of
enhancements and fixes over 3.5. Again, this is a snapshot and
contains some code under development.
Unfortunately, a complete release of version 4 is still delayed
because we have bugs in some new kernel code. However this bugged
code is #ifdef'd out in the 3.6 source, which will compile and run.
You can either compile "make abi" with -DACEDB3 or "make xace" with
-DACEDB4. Apart from the fact that abi contains the acembly code,
these are essentially identical (you can check this easily). The
acembly code will later be in version 4 under a -DACEMBLY
Both of us (Jean and Richard) have been working on the same source
code tree since the last workshop. This is what is in
test.source.3.6a.tar.gz. All the differences between version 2 and
version 3 have been either eliminated (including the map model
structures, which are now a direct development in the line of version
3), or have been removed from the code and are setable by object data
values (for some of the default display differences). There has been
no development of version 2 code for several months.
As from 3.5, the files tags.wrm and clases.wrm no longer exist. This
mean that you no longer have to worry about compatibility of tag
numbers, the same binary can run on databases using different
tags. Furthermore, to define a new class or tag you need only to
update the model.wrm file.
Changes in objects are now tracked both in the object (toggle
Timestamp in the tree display menu) and in KeySet class objects.
Sorted keysets are now cached for large classes, lengthening some save
operations, but dramatically shortening display of these classes.
The table maker now correctly accesses data matrices, can use
parameters, and use arithmetic filters.
The GRID display now uses the same tag2 system as the Map, using
Positive and Negative as the primary tags, and is configured using
View class objects.
The VMAP map display is as it always has been for version 3.
There is a new GMAP map display using Simon Kelley's new columns code.
This uses exactly the same models for positioning objects as VMAP.
However if you use the map data classes (2pt, Multipt and Df_dup)
these models have changed to use tag2 systems to allow data for
arbitrary objects. It too can be configured using View class objects.
Like other parts of this code, it is currently in test state.
Stan Letovsky's metabolic pathways code has been incorporated (display
type METAB), with minor edits and fixes.
In addition to the wspec/ directory, which is in a direct line with
previous version 3 releases, there is a wspec.4/ directory. The
models.wrm file here contains the new models for Grid, View, map data
and metabolic pathways classes. However it also contains changes to
the ?Sequence and ?Peptide models that are not reflected in 3.6 source
code, so do not try to use them yet. You are welcome to look at them
and comment. They should probably be the subject of a separate message.
The client server system has been significantly improved over 3.4
and is now starting to be used as a lab notebook in the Saint Louis
sequencing center. The server and the tace interface allow new
operations, in particular global editions on the active list and
tables with parameters (type ? on the command line).
To compile the client server, read the README file in the wrpc
directory. On Solaris only, one of the file generated by rpc must be
edited by hand. The server can handle several clients in parallel on
arbitrary platforms (Sun, Dec, Silicon, IBM have been tested). We
provide a simple perl interface. In Saint Louis we have client calls
embeded in tcl/tk scripts, these are avilable on request.
The sequence editor and assembly code, developed by Danielle, Jean and
Ulrich, is not yet complete, but is already capable of editing ABI
trace files and automatically assembling a cosmid into a few contigs
at the end of the shotgun phase. It can interact with xBap and with
Phil Green's codes. Test data files are available on request.
Richard and Jean
More information about the Acedb