So. Very novice. I added these rack tabs to a saddlebag support I’m working on. I placed them flush to the outside of the 3/8 - 0.049 tube. Perhaps I should have centered them. The TIG welds were bad. I ground them down, because I wasn’t certain I had good penetration on both parts. Grinding alleviated my worries, and so I put a fresh pass of ugly back on it. I’ll file and clean them up later. My question now is regarding the heal of the tab. I dremeled the tube to fit up to the tab fairly well, but I probably should have shaped the tab to match the tube instead. Regardless, that’s a gap, and since this will be under my saddle, it’ll be resisting side to side motion. I believe the smart solution is to weld that gap shut. Smarter would be to put a gusset over it. Less smart may be to try to fill it with silver. And short lived might be to fill it with paint… I do have an earlier version which I welded the same as this current state and it has not had any issues, so maybe the lateral loading isn’t as severe as I expect. Thoughts and guidance will be much appreciated.
It’ll be fine as it is, but I would weld it shut anyway. Doesn’t need a gusset. Make sure the tubes are vented somewhere or hot gas can blow back out of your weld right at the end just as you’re closing it up.
Maybe not the most “Instagram” welds but they look functional in these pictures so don’t worry about it!
The tabs on these things are much thicker than the tube so that they can support bolts without getting crushed. You don’t need massive side-to-side stiffness, because both sides are going to be bolted in place across the bike. The welds you have look at least twice the wall thickness of the tube, so, assuming they’re decent, if you were to break test this, I think the tube would fail before the welds broke. That’s the acid test for whether you’ve done enough, which I think you have. But I would still weld them just because reasons. Personal/aesthetic/OCD or something 
Thanks for input and feedback.
I wasn’t really trying to skip steps, but I was a little nervous about making it worse than it was. I need to have this done for a trip, and won’t have time to start again from scratch.
Anyway, the gaps filled in easily enough, though I wish I’d put the tabs on before bending the end so close together. It was a little difficult to find a shot I liked.
I don’t tig but welding across the top of the tab was my thought as well.
Can I ask why you didn’t just weld around the whole tab to start with?
Well, of course. I’m just very inexperienced and was lacking confidence to accomplish it correctly.
Welding along the straight edge was simple enough. But I was concerned that the gap and needing to follow the curve would cause me trouble, and I might end up making a mistake that would require starting over. So, really it was just a plea for someone to tell me it was fine as is.
I had already had enough trouble bending it with a Ridgid 606 bender. First it was ovalizing bends in thinner 0.028 and 0.035 4130. I have not had that issue with 304ss. Then, with the 0.049, I really attempted to get the bends sequenced properly and aligned planarly(?) to minimize strong arming it afterward. I had the bender in a vice, used the digital protractor on my phone, and was as careful and attentive as I could be. Then, I spent a ton of time strong arming it. And that was a lesson it how much more resilient the 4130 is. I really think my bends were better aligned when I eyeballed it.
Anyway, it’s rattle canned and on the bike now. I’ll snap a finished photo tonight.
I’ve often found that the Mark I eyeball turns out to be more reliable than trying to measure things. You can see particular things, like parallelism between two things that are close together for example, very accurately, often more so than the accuracy with which you can get your measurement tools properly aligned with the thing you’re trying to measure.
It looks like your rack has turned out very well. A lot of complex bends there!