Struct EnvParam

Source

pub struct EnvParam { /* private fields */ }

Expand description

One row of C’s ENV_PARAM table (envDefs.h:41-45): a parameter name and the default compiled in from configure/CONFIG_ENV.

Values of this type exist only in the generated super::env_table.

Implementations§

Source §

impl EnvParam

Source

pub const fn name(&self) -> &'static str

Source

pub const fn default_str(&self) -> &'static str

The compiled-in default string, exactly as it appears in C’s envData.c.

Source

pub fn get(&self) -> Option<String>

C envGetConfigParamPtr (envSubr.c:81-100): the environment string, falling back to the compiled default — with an empty result mapped to None.

The None/Some answer is a load-bearing presence test, not just an absence check: caservertask.c:491-508 uses it to pick which of two parameters to resolve, so FOO= must read as “not configured”.

Source

pub fn long(&self) -> Option<i64>

C envGetLongConfigParam (envSubr.c:303-320) — sscanf("%ld") over the resolved value, out of the first 127 bytes (all C’s char text[128] keeps).

None is C’s non-zero status. A set-but-unparseable value also prints C’s stderr line first; an unresolvable one (empty value and empty default) prints nothing, because envGetConfigParam returned NULL before sscanf ever ran.

Source

pub fn long_or_default(&self) -> i64

The C caller idiom for a long-valued parameter:

long i = 50;                            /* == the table default */
envGetLongConfigParam(&IOCSH_HISTSIZE, &i);

i.e. a rejected value leaves the compiled default standing. The 50 in C’s source is a hand-copy of the table; here it comes from the table.

Source

pub fn double(&self) -> Result<f64, EnvDoubleError>

C envGetDoubleConfigParam (envSubr.c:191-211) — epicsScanDouble over the resolved value.

An unset parameter resolves to its compiled default and parses, so Ok(30.0) for EPICS_CA_CONN_TMO on a clean shell. Err is C’s non-zero status: the value (or the default) is empty, or it did not scan — and in the latter case C’s stderr line has already been printed. C’s callers then print their own named diagnostic on top of it.

Source

pub fn bool(&self) -> Option<bool>

C envGetBoolConfigParam (envSubr.c:324-333): *pBool = epicsStrCaseCmp(text, "yes") == 0, i.e. true iff the resolved value is the word yes, any case. "1", "true", "on", "yes " are all false.

None is C’s -1 status — nothing to compare, so C leaves the caller’s variable untouched. The caller’s unwrap_or(..) is then C’s own initializer, not a second copy of the table default.

Source

pub fn inet_port(&self, fallback: u16) -> u16

C envGetInetPortConfigParam (envSubr.c:397-424).

fallback is the compiled port the C caller passes as defaultPort — for the EPICS_CAS_* overrides that is a different parameter’s default (caservertask.c:491-508 passes CA_SERVER_PORT), which is why it is an argument rather than self.default. Derive it with EnvParam::default_port so it still comes from the table.

Not resolvable, or no integer found: the “integer fetch failed” diagnostics, then fallback.
<= IPPORT_USERRESERVED (5000) or > USHRT_MAX: the “out of range” diagnostics, then fallback.

The diagnostics are byte-identical to compiled C, once per resolution — there is no dedup in C:

Unable to find an integer in EPICS_CA_SERVER_PORT=abc
EPICS Environment "EPICS_CA_SERVER_PORT" integer fetch failed
setting "EPICS_CA_SERVER_PORT" = 5064

EPICS Environment "EPICS_CA_SERVER_PORT" out of range
Setting "EPICS_CA_SERVER_PORT" = 5064

Source

pub const fn default_port(&self) -> u16

The compiled default read as a port, at compile time.

This is how a pub const CA_SERVER_PORT: u16 is derived from the table instead of hand-written next to it: a default that is not a valid port fails const-evaluation, so the constant cannot drift from CONFIG_ENV.

Source