Store defined symbols #220
Conversation
|
@tjj2017 : your thoughts on this one would be appreciated. |
| @@ -1151,6 +1152,20 @@ package body Tree_Walk is | |||
| Set_Identifier (Sym, Unique_Name (E)); | |||
| Set_Type (Sym, Symbol_Type); | |||
|
|
|||
| if not Global_Symbol_Table.Contains (Sym_Id) then | |||
There was a problem hiding this comment.
Basically, my understanding is this should only happen if we either didn't see the declaration of the type before somehow (which seems unlikely, I hope), or we couldn't parse the declaration (which would explain the many 'fixes' I suppose).
I'm still slightly skeptical this will handle all cases properly, but I'd suspect it's only going to break in cases which are otherwise broken anyway.
There was a problem hiding this comment.
I tend to agree with this analysis. I believe it is a general Ada rule that you cannot reference an entity until it has been at least incompletely defined.
|
It'd still be good to have a test case, though if my suspicion is correct it's actually not possible to construct a stable test case for this... |
There was a problem hiding this comment.
This looks very similar to the work I did on #238. I think in that piece the symbol registering is happening earlier (during the Do_Object_Declaration), because I had also found that some symbols that I was looking up where not in the symtab. Could it be that one of the two works is extraneous or do we need both? I will look into it perhaps a bit later and come back to you on that. Otherwise, LGTM
|
I'll rebase this PR as a new PR in #248. |
To update the symbol table if the identifier being defined is missing.
1: use wrapper in do_defining identifier 2: avoid duplicate call in do_full_object_declaration 3: update initial value if symbol is present.
xbauch
force-pushed
the
feature/store-defined-symbols
branch
from
86cc890 to
765978d
Compare
Fix PR #220 Regression - Urgent
during
Do_Defining_Identifier, in the symbol table ,if they are not present already.