DocsThere is no absolute requirement for a parser to deal with anything but the particular encoding syntax defined in this specification. However, it is possible that Compound Text may be extended in the future, and as such it may be desirable to construct the parser to handle 2022/6429 syntax more generally.
There are two general formats covering all control sequences that are expected to appear in extensions:
01/11 {I} F
For this format, I is always in the range 02/00 to 02/15, and F is always in the range 03/00 to 07/14.
09/11 {P} {I} F
For this format, P is always in the range 03/00 to 03/15, I is always in the range 02/00 to 02/15, and F is always in the range 04/00 to 07/14.
In addition, new (singleton) control characters (in the C0 and C1 ranges) might be defined in the future.
Finally, new kinds of "segments" might be defined in the future using syntax similar to extended segments:
01/11 02/05 02/15 F M L
For this format, F is in the range 03/05 to 3/15. M and L are as defined in extended segments. Such a segment will always be followed by the number of octets defined by M and L. These octets can have arbitrary values and need not follow the internal structure defined for current extended segments.
If extensions to this specification are defined in the future, then any string incorporating instances of such extensions must start with one of the following control sequences:
| 01/11 02/03 V 03/00 | ignoring extensions is OK |
| 01/11 02/03 V 03/01 | ignoring extensions is not OK |
In either case, V is in the range 02/00 to 02/15 and indicates the major version minus one of the specification being used. These version control sequences are for use by clients that implement earlier versions, but have implemented a general parser. The first control sequence indicates that it is acceptable to ignore all extension control sequences; no mandatory information will be lost in the process. The second control sequence indicates that it is unacceptable to ignore any extension control sequences; mandatory information would be lost in the process. In general, it will be up to the client generating the Compound Text to decide which control sequence to use.