Cache register names

[quanzhuo]

Cache register names

[quanzhuo]

Sorry that two unrelated commits appear in the same PR.
The first commit is for caching register names to avoid requesting them every time.
The second commit removes the stopAtEntry property (only for attach requests), as the stopAtEntry property is meaningless for attach-type requests.
@WebFreak001 Hi, boss could you please take a while and look at this ?

[GitMensch]

stopAtEntry should still work and lead to "stop at attach" (but that may be automatic) and if set to an entry lead to a temporary breakpoint at the given function name, no? It does look like the code you've adjusted would do that...

[GitMensch]

the idea is good, but we need to store the regnames per inferior (you may switch between multiple ones via UI), not globally, as the inferiors can have different registers (example: server in 64 bit and client in 32 bit (or vice versa), attached to both)

[GitMensch]

please check the scope of the caching and add a ChangeLog entry as well (one for each commit, but I'm not convinced yet that the second commit should be merged - maybe focus on the register caching here and discuss dropping the setting [which also needs a ChangeLog entry] separately)

[WebFreak001]

rest LGTM, just GitMensch's comment

[brownts]

The second commit removes the stopAtEntry property (only for attach requests), as the stopAtEntry property is meaningless for attach-type requests.

Please don't remove this functionality, it is not meaningless. I've used this for many years and others have as well. Was there some problem you were trying to solve by removing this?

In the original PR thread (see here), I described how this works for an "attach" when used in combination with gdbserver where the application does not start running immediately. In that context it is valid to run the application to the entry point.

[quanzhuo]

Close this PR first. When I'm ready, I'll submit a new one.
Thanks all.