Normally, dynamic abbrev expansion ignores case when searching for expansions. That is, the expansion need not agree in case with the word you are expanding.
This feature is controlled by the variable
dabbrev-case-fold-search. If it is
t, case is ignored
in this search; if it is
nil, the word and the expansion must
match in case. If the value is
case-fold-search (the default),
then the variable
case-fold-search controls whether to ignore
case while searching for expansions (see Lax Search).
Normally, dynamic abbrev expansion preserves the case pattern of the dynamic abbrev you are expanding, by converting the expansion to that case pattern.
dabbrev-case-replace controls whether to
preserve the case pattern of the dynamic abbrev. If it is
the dynamic abbrev’s case pattern is preserved in most cases; if it is
nil, the expansion is always copied verbatim. If the value is
case-replace (the default), then the variable
case-replace controls whether to copy the expansion verbatim
(see Replacement and Lax Matches).
However, if the expansion contains a complex mixed case pattern, and
the dynamic abbrev matches this pattern as far as it goes, then the
expansion is always copied verbatim, regardless of those variables.
Thus, for example, if the buffer contains
variableWithSillyCasePattern, and you type v a M-/, it
copies the expansion verbatim including its case pattern.
dabbrev-abbrev-char-regexp, if non-
controls which characters are considered part of a word, for dynamic expansion
purposes. The regular expression must match just one character, never
two or more. The same regular expression also determines which
characters are part of an expansion. The (default) value
has a special meaning: dynamic abbrevs are made of word characters,
but expansions are made of word and symbol characters.
In shell scripts and makefiles, a variable name is sometimes prefixed
with ‘$’ and sometimes not. Major modes for this kind of text can
customize dynamic abbrev expansion to handle optional prefixes by setting
dabbrev-abbrev-skip-leading-regexp. Its value
should be a regular expression that matches the optional prefix that
dynamic abbrev expression should ignore. The default is
which means no characters should be skipped.