Dubious logical expression in rsslReactor.c

[DuncanSands]

When compiling RTSDK-2.3.2.L1.linux.rrg, g++ 15.2.0 warns about the code at Eta/Impl/Reactor/rsslReactor.c line 6827:

                                                                    /* If this isn't the first warm standby group (or preferred) and there is a login callback */
                                                                    if (!pWarmStandByHandlerImpl->isInitialConnection == RSSL_TRUE
                                                                            && pReactorWarmStandbyServerInfo->reactorConnectInfoImpl.base.enableSessionManagement == RSSL_FALSE
                                                                            && pStandByReactorChannel->channelRole.ommConsumerRole.pLoginRequestList[pStandByReactorChannel->currentConnectionOpts->base.loginReqIndex]->pLoginRenewalEventCallback != NULL)
                                                                    {

And indeed is that initial exclamation mark supposed to apply to all of the rest, or only to pWarmStandByHandlerImpl->isInitialConnection? g++ suggests adding brackets to clarify.

[L-Karchevska]

@DuncanSands Thank you for pointing this out, we will address this issue.

[L-Karchevska]

@DuncanSands could you specify the platform you are using to reproduce the issue?

[DuncanSands]

This is when building with gcc 14.3.1 and warning option -Wlogical-not-parentheses (not g++, sorry).
Platform is linux (Ubuntu 25.10)

I think the simplest thing to do is to add parentheses that clarify the intent. For example, this silences the warning:

--- a/External/Elektron/RTSDK-2.3.2.L1.linux.rrg/Eta/Impl/Reactor/rsslReactor.c
+++ b/External/Elektron/RTSDK-2.3.2.L1.linux.rrg/Eta/Impl/Reactor/rsslReactor.c
@@ -6824,7 +6824,7 @@ static RsslRet _reactorDispatchEventFromQueue(RsslReactorImpl pReactorImpl, Rss
if (pStandByReactorChannel->channelRole.ommConsumerRole.pLoginRequestList)
{
/ If this isn't the first warm standby group (or preferred) and there is a login callback */

•                                                                   if (!pWarmStandByHandlerImpl->isInitialConnection == RSSL_TRUE
  

•                                                                   if ((!pWarmStandByHandlerImpl->isInitialConnection) == RSSL_TRUE
                                                                            && pReactorWarmStandbyServerInfo->reactorConnectInfoImpl.base.enableSessionManagement == RSSL_FALSE
                                                                            && pStandByReactorChannel->channelRole.ommConsumerRole.pLoginRequestList[pStandByReactorChannel->currentConnectionOpts->base.loginReqIndex]->pLoginRenewalEventCallback != NULL)
                                                                    {
  

as does this:

--- a/External/Elektron/RTSDK-2.3.2.L1.linux.rrg/Eta/Impl/Reactor/rsslReactor.c
+++ b/External/Elektron/RTSDK-2.3.2.L1.linux.rrg/Eta/Impl/Reactor/rsslReactor.c
@@ -6824,9 +6824,9 @@ static RsslRet _reactorDispatchEventFromQueue(RsslReactorImpl pReactorImpl, Rss
if (pStandByReactorChannel->channelRole.ommConsumerRole.pLoginRequestList)
{
/ If this isn't the first warm standby group (or preferred) and there is a login callback */

•                                                                   if (!pWarmStandByHandlerImpl->isInitialConnection == RSSL_TRUE
  

•                                                                   if (!(pWarmStandByHandlerImpl->isInitialConnection == RSSL_TRUE
                                                                            && pReactorWarmStandbyServerInfo->reactorConnectInfoImpl.base.enableSessionManagement == RSSL_FALSE
  

•                                                                           && pStandByReactorChannel->channelRole.ommConsumerRole.pLoginRequestList[pStandByReactorChannel->currentConnectionOpts->base.loginReqIndex]->pLoginRenewalEventCallback != NULL)
  

•                                                                           && pStandByReactorChannel->channelRole.ommConsumerRole.pLoginRequestList[pStandByReactorChannel->currentConnectionOpts->base.loginReqIndex]->pLoginRenewalEventCallback != NULL))
                                                                    {
                                                                            RsslReactorLoginRenewalEvent loginEvent;
                                                                            rsslClearReactorLoginRenewalEvent(&loginEvent);
  

Which one is it supposed to be?

[fogol]

@DuncanSands Does it imply that you are building with gcc with extra warning options enabled, but not default gcc options set? In this particular case, did you explicitly provide warning option -Wlogical-not-parentheses to the build options? If you did not provide it explicitly, where did this particular one and other options come from?

[DuncanSands]

I'm using -Wall, which turns on -Wlogical-not-parentheses. In practice I turn off many warnings, because otherwise it is out of control:

C and C++:

--  Warnings that don't seem worth us fixing.
"-Wno-misleading-indentation",
"-Wno-sign-compare",
"-Wno-switch",
"-Wno-unused-function",
"-Wno-unused-value",
"-Wno-unused-variable",

-- TODO: report or fix and send upstream.
"-Wno-array-bounds",
"-Wno-comment",
"-Wno-logical-not-parentheses",
"-Wno-maybe-uninitialized",
"-Wno-missing-braces",
"-Wno-stringop-truncation",
"-Wno-parentheses"

C specific:

--  Warnings that don't seem worth us fixing.
"-Wno-pointer-sign"

C++ specific:

--  Warnings that don't seem worth us fixing.
"-Wno-char-subscripts",
"-Wno-reorder",

-- TODO: report or fix and send upstream.
"-Wno-overloaded-virtual",
"-Wno-register",
"-Wno-restrict",
"-Wno-tautological-compare"

It would be great if you compiled yourselves with g++/gcc and -Wall and tried to eliminate as many warnings as you reasonably can. Some of them seem genuinely worrying.

Note that I am unable to use cmake+make due to the way our internal build system works. Perhaps you suppress warnings there or never turn them on in the first place.

[fogol]

@DuncanSands Thank you for submitting your issue. The fix will be included in a future release.