That's succinct and to the point. I would elaborate a bit more:

The rules concerning the definition and use of aliases are somewhat confusing. Essentially an alias command needs to have been executed before code using it is parsed. Further, keep in mind that bash always reads and parses at least one complete line of input (which may include several commands including alias commands), but also any number of lines that make up a compound command (including function-definitions), before executing any of the commands. Overlooking this will lead to "command not found" errors at points of use within a complex command even after the alias definition.

(Made the mistak eof just replyinf to the notification mail first)
Ha, ha, that's a good one😃 I of course know that passage from the manual - but I am bright enough to be willing to admit that I did not get it - for good reason. The point is that the reference to "line" is misleading: It should be clarified that a a complete command is meant - which could be simple (only then generally being a "line"), a pipeline, a list or a compound command (which I understand to include beyond "if"-blocks also function-definitions).

Thanks for the explanation, I was assuming that the alias is "known" (and can be used) once it is parsed - ahead of code like functions using it - whereas I now understand that the alias needs to be executed before code using them is parsed. I had already segregated alias definitions to their own files (kind of like C-headers with definitions) and had meanwhile also sucessfully changed the logic of sourcing - I source conditionally but in a command list (like "[ "$lib" ] || source lib.sh") rather than in an if-block.

Maybe worth a comment in the manual somewhere, e.g. on compound commands or syntax in general under parsing vs. execution ...

Regards, Martin

In trying to use a popular C-idiom in a library of shell-script(lets) I conditionalized code on the file not already being sourced - and then found that function-definitions no longer recognized valid alias-definitions (coming from other files, but that is probably immaterial), despite doing "shopt -s expand_aliases" - in fact the alias no longer works there at all.

Consider an example using the attachments: A "library" test-source-alias.sh sources test-source-alias-aliases.sh (which defines an alias test_alias) and then defines a function test_function. This fails when the function-definition (or even a call of test_alias) is in an if-block - but works outside. Running the example follows ...
[martin@MartinsDell ~]$ set -x
++ printf '\033]0;%s@%s:%s\007' martin MartinsDell '~'
[martin@MartinsDell ~]$ LC_ALL=C
+ LC_ALL=C
++ printf '\033]0;%s@%s:%s\007' martin MartinsDell '~'
[martin@MartinsDell ~]$ source test-source-alias.sh
+ source test-source-alias.sh
++ shopt -s expand_aliases
++ '[' '!' '' ']'
++ test_source_alias=true
++ source test-source-alias-aliases.sh
+++ '[' '!' '' ']'
+++ test_source_alias_aliases=true
+++ alias 'test_alias=echo test_alias called'
++ alias
alias e='emacsclient -c'
alias ll='ls -lF --show-control-chars'
alias ls='ls -F --color=auto --show-control-chars'
alias nosleep='systemd-inhibit'
alias test_alias='echo test_alias called'
++ shopt
autocd          off
cdable_vars     off
cdspell         off
checkhash       off
checkjobs       off
checkwinsize    on
cmdhist         on
compat31        off
compat32        off
compat40        off
compat41        off
compat42        off
compat43        off
complete_fullquote      on
direxpand       off
dirspell        off
dotglob         off
execfail        off
expand_aliases  on
extdebug        off
extglob         off
extquote        on
failglob        off
force_fignore   on
globasciiranges off
globstar        off
gnu_errfmt      off
histappend      off
histreedit      off
histverify      off
hostcomplete    on
huponexit       off
inherit_errexit off
interactive_comments    on
lastpipe        off
lithist         off
login_shell     off
mailwarn        off
no_empty_cmd_completion off
nocaseglob      off
nocasematch     off
nullglob        off
progcomp        on
promptvars      on
restricted_shell        off
shift_verbose   off
sourcepath      on
xpg_echo        off
++ test_alias
bash: test_alias: command not found
++ test_function
++ test_alias
bash: test_alias: command not found
++ alias
alias e='emacsclient -c'
alias ll='ls -lF --show-control-chars'
alias ls='ls -F --color=auto --show-control-chars'
alias nosleep='systemd-inhibit'
alias test_alias='echo test_alias called'
++ shopt
autocd          off
cdable_vars     off
cdspell         off
checkhash       off
checkjobs       off
checkwinsize    on
cmdhist         on
compat31        off
compat32        off
compat40        off
compat41        off
compat42        off
compat43        off
complete_fullquote      on
direxpand       off
dirspell        off
dotglob         off
execfail        off
expand_aliases  on
extdebug        off
extglob         off
extquote        on
failglob        off
force_fignore   on
globasciiranges off
globstar        off
gnu_errfmt      off
histappend      off
histreedit      off
histverify      off
hostcomplete    on
huponexit       off
inherit_errexit off
interactive_comments    on
lastpipe        off
lithist         off
login_shell     off
mailwarn        off
no_empty_cmd_completion off
nocaseglob      off
nocasematch     off
nullglob        off
progcomp        on
promptvars      on
restricted_shell        off
shift_verbose   off
sourcepath      on
xpg_echo        off
++ echo test_alias called
test_alias called
++ test_function
++ echo test_alias called
test_alias called
++ printf '\033]0;%s@%s:%s\007' martin MartinsDell '~'
[martin@MartinsDell ~]$

Am I missing something or is this intended behavioud?

Regards, Martin