#317 closed defect (fixed)
clarification of SDRS section 5.3.4 "Configuration Files"
| Reported by: | jhoblitt | Owned by: | jhoblitt |
|---|---|---|---|
| Priority: | high | Milestone: | |
| Component: | IPP SDRS | Version: | unspecified |
| Severity: | normal | Keywords: | IPP-doc |
| Cc: |
Description
- How are NULL, NAN, or INF values declared? I have a particular need for NULL
values.
- Are "multiple symbols" values required to all be of the same type (the SDRS
doesn't state one way or the other)? Is each declaration for a "multiple
symbol" intended to be a separate psMetadataItem with the same name or a nested
psMetadata structure?
- Are all statements terminated by a new line or are they allowed to span
multiple lines (spanning multiple lines causes trouble with the comment line
syntax)? Is '\n' the only valid new line or is '\r\n' supported?
- I assume that only ASCII is supported but it would be nice if this was noted.
- Can 'METADATA's be nested? e.g.
--
CELL METADATA
...
OTKERNEL METADATA
...
END
...
END
--
- Is both the 'TYPE' syntax and the 'METADATA' syntax required? Both result in
the same datastrucutre.
- Are complex type constants going to be required?
- The grammar in appendix A supports all of C99's constant declarations but many
of these are redundant because the type information is avaiable. Is it possible
to (at least for now) to reduced this to some subset (do we really need octal
constants?)?
Change History (11)
comment:1 by , 21 years ago
comment:2 by , 21 years ago
| bug_group: | PSLib? → IPP-doc? |
|---|---|
| Component: | astro → IPP SDRS |
| Keywords: | IPP-doc added; PSLib removed |
| product: | PSLib → IPP-doc |
| Status: | new → assigned |
comment:3 by , 21 years ago
Are [nested/unested] metadata entries allowed to be empty? If they are allowed,
should a warning be generated? Should they be removed from the parse tree?
comment:4 by , 21 years ago
Are comments allowed as part of a metadata declaration? If so, are they ignored
or do they become part of the parent MetadataItem? Eg.
CELL METADATA # foo
Are comments allowed after a metadata closing? Eg.
END # foo
Are types (STR|METADATA|etc.) case sensitive?
comment:5 by , 21 years ago
Are "multiple symbols" scoped? ie. Are they visible in nested metadata
statements at a lower "level"? Or, if one is declared in a nested metadata is
it then visible at a higher level?
comment:6 by , 21 years ago
Since this config language doesn't have conditionals, blocks, constants, or a
way to include other files... does it make sense for the parser to first run the
C preproccesor?
comment:7 by , 21 years ago
| Owner: | changed from to |
|---|---|
| Status: | assigned → new |
If all of your concerns in this bug have been resolved in the latest SDRS,
please set to resolve.
If any are outstanding, please note or re-file the individual problems.
comment:8 by , 21 years ago
| blocked: | → 391 |
|---|
comment:9 by , 21 years ago
| blocked: | 391 |
|---|---|
| dependson: | → 391 |
comment:10 by , 21 years ago
| Status: | new → assigned |
|---|

A date & time type would be useful as well. I'd like to propose one of the
formats below.
timestamp DATETIME 2005-03-02 16:16:00
This format isn't ISO8601 comptible but it does have the advantage of being
MySQL's native type.
An alternative format could be:
timestamp DATETIME 2005-03-02T16:24:00.0Z
This is part of ISO8601 standard and is unambiguous as to which timezone your
in. If you think that subsecond resolution is import you might as well go with
the later format as MySQL can't understand subsecond time, so the format will
have to be mangled into a real number anyways...