Message ID | 1460069165-5638-1-git-send-email-nhofmeyr@sysmocom.de |
---|---|
State | Not Applicable |
Headers | show |
Hi Max, I disagree with this patch. We don't want to make all warnings quiet at all cost. If that was the goal, we could simply disable compiler warnings and go ahead. The point is that the SUA code is incomplete, and I prefer to have a warning about it until it gets properly fixed. This: On Fri, Apr 08, 2016 at 12:46:05AM +0200, Neels Hofmeyr wrote: > - cause = xua_msg_get_u32(xua, SUA_IEI_CAUSE); > + /* TODO evaluate cause: > + * cause = xua_msg_get_u32(xua, SUA_IEI_CAUSE); */ will just mean that we all forget about having to fix it.
On Tue, Apr 12, 2016 at 12:23:52AM +0200, Harald Welte wrote: > Hi Max, It was me, not Max :) > will just mean that we all forget about having to fix it. Maybe instead of numerous unused variables, we should have a single #warning then. Every warning hinders my dev cycle. For pretty much every compilation, I have to parse/browse the list of warnings, *all* the time. Even a #warning would stare me in the face for every build. There are a couple of those already. So I'd even prefer a TODO file in / , admitted that these kind of files tend to go outdated. Also, having warnings around makes it harder to spot new warnings I've introduced. I see some warnings scroll by, then assume they're just the old ones, or have to scroll around to check every time. I'd prefer a way that I don't have to check for and read the same old warnings dozens of times every day, while not having the time to investigate/fix anyway... I unnecessarily lose time on those. A workaround for me would be to eradicate the warnings on our iu branches but not commit them to master. ~Neels
Hi Neels, On Tue, Apr 12, 2016 at 01:17:34PM +0200, Neels Hofmeyr wrote: > On Tue, Apr 12, 2016 at 12:23:52AM +0200, Harald Welte wrote: > > Hi Max, > > It was me, not Max :) Sorry. > > will just mean that we all forget about having to fix it. > > Maybe instead of numerous unused variables, we should have a single > #warning then. Every warning hinders my dev cycle. For pretty much every > compilation, I have to parse/browse the list of warnings, *all* the time. Please let me know after how many compilatiosn the trade-off woudl trip towards actually addressing those warnings in fixing the related code. > Even a #warning would stare me in the face for every build. There are a > couple of those already. Well, if everyone every so often would address one of them, it wouldn't be so bad :)
diff --git a/src/sua.c b/src/sua.c index 6e187d8..df60c13 100644 --- a/src/sua.c +++ b/src/sua.c @@ -793,7 +793,6 @@ static int sua_rx_core(struct osmo_sua_link *link, struct xua_msg *xua) struct xua_msg_part *data_ie = xua_msg_find_tag(xua, SUA_IEI_DATA); struct msgb *upmsg; struct sua_connection *conn; - uint8_t *cur; /* fill conn */ conn = conn_create(link); @@ -926,7 +925,8 @@ static int sua_rx_coref(struct osmo_sua_link *link, struct xua_msg *xua) memcpy(¶m->calling_addr, &conn->calling_addr, sizeof(param->calling_addr)); //param->in_sequence_control; - cause = xua_msg_get_u32(xua, SUA_IEI_CAUSE); + /* TODO evaluate cause: + * cause = xua_msg_get_u32(xua, SUA_IEI_CAUSE); */ /* optional: src addr */ /* optional: dest addr */ param->importance = xua_msg_get_u32(xua, SUA_IEI_IMPORTANCE);