Fabio/ddclient
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Add/update TODO comments for problematic bits of code

Browse source
This commit is contained in:
Richard Hansen 2024-08-22 02:22:25 -04:00
parent cf54da50e4
commit 4b5f28b2f0
1 changed files with 82 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

99
ddclient.in
View file

@ -138,11 +138,31 @@ $ENV{'PATH'} = (exists($ENV{PATH}) ? "$ENV{PATH}:" : "") . "/sbin:/usr/sbin:/bin
our %globals; our %globals;
our %config; our %config;
# %recap holds details about recent updates (and attempts) that are needed to implement various # `%recap` holds details about recent updates (and attempts) that are needed to implement various
# service-specific and protocol-independent mechanisms such as `min-interval`. This data is # service-specific and protocol-independent mechanisms such as `min-interval`. This data is
# persisted in the cache file (`--cache`) so that it survives ddclient restarts. This hash maps a # persisted in the cache file (`--cache`) so that it survives ddclient restarts. This hash maps a
# hostname to a hashref containing those protocol variables that have their `recap` property set to # hostname to a hashref with entries that map variable names to values. Only entries for the
# true. # host's "recap variables" -- the host's protocol's variables with a truthy `recap` property -- are
# included.
#
# `%config` is the source of truth for recap variable values. The recap variable values in
# `%config` are initialized with values from the cache file (via `%recap`). After initialization,
# any change to a recap variable's value is made to `%config` and later copied to `%recap` before
# being written to the cache file. Changes made directly to `%recap` are erroneous because they
# will be overwritten by the value in `%config`.
#
# There are two classes of recap variables:
# * "Status" variables: These track update success/failure, the IP address of the last successful
# update, etc. These do not hold configuration data. These should always be kept in sync with
# `%config`.
# * "Configuration change detection" variables: These are used to force an update if the value in
# the same-named entry in `%config` has changed since the previous update attempt. The value
# stored in `%config` is the desired setting; the value in `%recap` is the desired setting as
# it was just before the previous update attempt. Values are synchronized from `%config` to
# `%recap` during each update attempt.
#
# The set of config change detection variables is currently hard-coded; all other recap variables
# are assumed to be status variables.
# #
# A note about terminology: This was previously named `%cache`, but "cache" implies that the # A note about terminology: This was previously named `%cache`, but "cache" implies that the
# purpose is to reduce the cost or latency of data retrieval or computation, and that deletion only # purpose is to reduce the cost or latency of data retrieval or computation, and that deletion only
@ -642,9 +662,12 @@ our %variables = (
'ip' => setv(T_IP, 0, 0, undef, undef), 'ip' => setv(T_IP, 0, 0, undef, undef),
# As a recap value, this is the IPv4 address most recently saved at the DDNS service. As a # As a recap value, this is the IPv4 address most recently saved at the DDNS service. As a
# setting, this is the desired IPv4 address that should be saved at the DDNS service. # setting, this is the desired IPv4 address that should be saved at the DDNS service.
# Unfortunately, these two meanings are conflated, causing the bug "skipped: IP address was # TODO: The use of `ipv4` as a recap status variable is supposed to be independent of the
# already set to a.b.c.d" when the IP was never set to a.b.c.d. # use of `ipv4` as a configuration setting. Unfortunately, because the `%config` value is
# TODO: Move the recap value elsewhere to fix the bug. # synced to `%recap`, the two uses conflict which causes the bug "skipped: IP address was
# already set to a.b.c.d" when the IP was never set to a.b.c.d. Rename the `%recap` status
# variable to something like `saved-ipv4` to avoid the collision (and confusion) with the
# `%config` setting variable.
'ipv4' => setv(T_IPV4, 0, 1, undef, undef), 'ipv4' => setv(T_IPV4, 0, 1, undef, undef),
# As `ipv4`, but for an IPv6 address. # As `ipv4`, but for an IPv6 address.
'ipv6' => setv(T_IPV6, 0, 1, undef, undef), 'ipv6' => setv(T_IPV6, 0, 1, undef, undef),
@ -1539,8 +1562,17 @@ sub write_recap {
# nic_updateable (called from update_nics for each host) sets `$config{$h}{update}` # nic_updateable (called from update_nics for each host) sets `$config{$h}{update}`
# according to whether an update was attempted. # according to whether an update was attempted.
# #
# TODO: Why are different variables saved to the recap depending on whether an update was # TODO: The "configuration change detection" variables are also syncronized if `$recap{$h}`
# attempted or not? Shouldn't the same variables be saved every time? # doesn't exist. Why is this done? Perhaps the intention was to prevent the
# force-if-config-changed logic from triggering the very first cycle (when there is no
# previous update attempt). However:
# * There's a bug: This function isn't called until after `update_nics` has already
# checked the force-if-config-changed condition, so syncing the values here is too late
# to suppress the first cycle's force-if-config-changed logic.
# * It's unnecessary: If there have not been any update attempts, an update attempt
# should be made anyway, and already is. (`update_nics` already forces an update if
# `$recap{$h}` doesn't exist, and that check is done before the force-if-config-changed
# check.)
if (!exists $recap{$h} || $config{$h}{'update'}) { if (!exists $recap{$h} || $config{$h}{'update'}) {
my $vars = $protocols{opt('protocol', $h)}{variables}; my $vars = $protocols{opt('protocol', $h)}{variables};
for my $v (keys(%$vars)) { for my $v (keys(%$vars)) {
@ -1548,6 +1580,8 @@ sub write_recap {
$recap{$h}{$v} = opt($v, $h); $recap{$h}{$v} = opt($v, $h);
} }
} else { } else {
# TODO: Why aren't all "status" variables included in this update? (missing: mtime
# ipv4 ipv6 warned-min-interval warned-min-error-interval)
for my $v (qw(atime wtime status-ipv4 status-ipv6)) { for my $v (qw(atime wtime status-ipv4 status-ipv6)) {
$recap{$h}{$v} = opt($v, $h); $recap{$h}{$v} = opt($v, $h);
} }
@ -1604,16 +1638,29 @@ sub read_recap {
%opt = (); %opt = ();
$saved_recap = _read_config($config, $globals, "##\\s*$program-$version\\s*", $file); $saved_recap = _read_config($config, $globals, "##\\s*$program-$version\\s*", $file);
%opt = %saved; %opt = %saved;
# TODO: Why is this iterating over `keys(%recap)`? Shouldn't it iterate over `keys(%config)`
# instead?
for my $h (keys(%recap)) { for my $h (keys(%recap)) {
# TODO: `$config->{$h}` is guaranteed to exist because `$config` is `\%recap` and we're
# iterating over `keys(%recap)`.
next if !exists($config->{$h}); next if !exists($config->{$h});
# TODO: Why is this limited to this set of variables? Why not copy every recap var defined # Only status variables are copied from `%recap` into `%config` because the recap should
# for the host's protocol? # not change the configuration. See the documenting comments for `%recap`.
# TODO: Why aren't all status variables included in this update? (missing:
# warned-min-interval warned-min-error-interval)
# TODO: Hard-coding this list impairs readability and maintainability. In particular,
# different pieces of code need to know the list of status variables, and keeping them all
# in sync is error-prone (DRY). Also, different protocols might need different status
# variables.
for (qw(atime mtime wtime ip ipv4 ipv6 status-ipv4 status-ipv6)) { for (qw(atime mtime wtime ip ipv4 ipv6 status-ipv4 status-ipv6)) {
# TODO: Isn't $config equal to \%recap here? If so, this is a no-op. What was the # TODO: This is a no-op because `$config` is `\%recap`. I believe the original intent
# original intention behind this? To copy %recap values into %config? If so, is it # was to copy values that convey update status from `%recap` into `%config` so that
# better to just delete this and live with the current behavior (which doesn't seem to # ddclient can be restarted and resume where it left off and not violate
# be causing users any problems) or to "fix" it to match the original intention, which # `min-interval`, `min-error-interval`, etc. For the short term, this should copy the
# might introduce a bug? # values into `%config`, not `%recap`. In the long term, nothing should be copied from
# `%recap` into `%config` because `%config` is not the semantically appropriate place
# to record update status. Code that currently reads/sets `$config{$h}{'status-ipv4'}`
# (and friends) should instead access `$recap{$h}{'status-ipv4'}` directly.
$config->{$h}{$_} = $recap{$h}{$_} if exists $recap{$h}{$_}; $config->{$h}{$_} = $recap{$h}{$_} if exists $recap{$h}{$_};
} }
} }
@ -1812,13 +1859,29 @@ sub _read_config {
## verify that keywords are valid...and check the value ## verify that keywords are valid...and check the value
for my $k (keys %locals) { for my $k (keys %locals) {
$locals{$k} = $passwords{$k} if defined $passwords{$k}; $locals{$k} = $passwords{$k} if defined $passwords{$k};
# TODO: The checks below are incorrect for a few reasons:
#
# * It is not protocol-aware. Different protocols can have different sets of
# variables, with different normalization and validation behaviors.
# * It does not check for missing required values. Note that a later line or a
# command-line argument might define a missing required value.
# * A later line or command-line argument might override an invalid value, changing
# it to valid.
#
# Fixing this is not simple. Values should be checked and normalized after processing
# the entire file and command-line arguments, but then we lose line number context.
# The line number could be recorded along with each variable's value to provide context
# in case validation fails, but that adds considerable complexity. Fortunately, a
# variable's type is unlikely to change even if the protocol changes
# (`$variables{merged}{$var}{type}` will likely equal
# `$protocols{$proto}{variables}{$var}{type}` for each variable `$var` for each
# protocol `$proto`), so normalizing and validating values on a line-by-line basis is
# likely to be safe.
if (!exists $variables{'merged'}{$k}) { if (!exists $variables{'merged'}{$k}) {
warning("unrecognized keyword '%s' (ignored)", $k); warning("unrecognized keyword '%s' (ignored)", $k);
delete $locals{$k}; delete $locals{$k};
next; next;
} }
# TODO: This might grab an arbitrary protocol-specific variable definition, which could
# cause surprising behavior.
my $def = $variables{'merged'}{$k}; my $def = $variables{'merged'}{$k};
if (!eval { $locals{$k} = check_value($locals{$k}, $def); 1; }) { if (!eval { $locals{$k} = check_value($locals{$k}, $def); 1; }) {
warning("invalid variable value '$k=$locals{$k}': $@"); warning("invalid variable value '$k=$locals{$k}': $@");
@ -3549,6 +3612,8 @@ sub nic_updateable {
$config{$host}{'update'} = 1; $config{$host}{'update'} = 1;
$config{$host}{'atime'} = $now; $config{$host}{'atime'} = $now;
delete($config{$host}{$_}) for qw(wtime warned-min-interval warned-min-error-interval); delete($config{$host}{$_}) for qw(wtime warned-min-interval warned-min-error-interval);
# TODO: Why aren't all of the above "status" variables (the ones deleted from `%config`)
# also removed from `%recap`? (missing: wtime status status-ipv4 status-ipv6)
delete $recap{$host}{'warned-min-interval'}; delete $recap{$host}{'warned-min-interval'};
delete $recap{$host}{'warned-min-error-interval'}; delete $recap{$host}{'warned-min-error-interval'};
} else { } else {