Visual D dmdserver (page 2)

On Saturday, 15 January 2022 at 09:08:32 UTC, Rainer Schuetze wrote: > > > On 05/01/2022 10:23, frame wrote: >> [...] > > So let's give it another try: you can find the symbols of the new release here: https://github.com/dlang/visuald/releases/tag/v1.2.0 Thanks, will try later 👍

On 12/02/2022 09:09, frame wrote: > On Saturday, 15 January 2022 at 09:08:32 UTC, Rainer Schuetze wrote: > >> So let's give it another try: you can find the symbols of the new release here: https://github.com/dlang/visuald/releases/tag/v1.2.0 > > Mostly dmdserver crashes if I edit a file outside Visual Studio - maybe it can just check against the syntax before actually loading the changes? > > I loaded the dmdserver.pdb 1.2.0 and found an access violation in dmd's typesemd.d but this somehow doesn't match with the source: > > Frame index: > > #0 dmdserver.exe!dmd.typesem.typeSemantic:2223 > points to a function call > > but > > #1 dmdserver.exe!dmd.typesem.typeSemantic.visitDelegate:1672 > points to a '}' > > > #2 dmdserver.exe!dmd.typesem.typeSemantic:2233 > points to a '}' too > > > #3 dmdserver.exe!dmd.expressionsem.ExpressionSemanticVisitor.visit:4100 > points to a comment > > So I think there is mismatch with the dmd sources and the debug information? - I'm using 2098.1 > The dmdserver uses this fork of dmd: https://github.com/rainers/dmd/tree/dmdserver The call stack looks like an incomplete delegate type with the function type missing. Maybe this has to do with the server not stopping after failing to parse the source, so you can still have partial semantic analysis. In some cases, the origial parser inserts null pointers into the syntax tree, but I've tried to replace this with error objects. It is possible that new code in dmd introduced more instances of this. It would be good if you could provide an example source that reproduces the problem.

February 14, 2022

Re: Visual D dmdserver

Posted by frame
in reply to Rainer Schuetze

Permalink

frame

Posted in reply to Rainer Schuetze

Permalink

On Sunday, 13 February 2022 at 17:26:49 UTC, Rainer Schuetze wrote:

The dmdserver uses this fork of dmd: https://github.com/rainers/dmd/tree/dmdserver

The call stack looks like an incomplete delegate type with the function type missing.

Maybe this has to do with the server not stopping after failing to parse the source, so you can still have partial semantic analysis. In some cases, the origial parser inserts null pointers into the syntax tree, but I've tried to replace this with error objects. It is possible that new code in dmd introduced more instances of this.

It would be good if you could provide an example source that reproduces the problem.

Yes, makes more sense with the right sources now.
The error may occurred as I re-organized my modules and a type became unavailable(?)

The typeSemantic function is called with a null-argument indeed:

type = null
loc.filename = "..\dmd2\src\phobos\std\array.d"

Something related to Appender.put!, to precise this line (3513):

auto bigData = (() @trusted => _data.arr.ptr[0 .. len + 1])();
               ^

called from ForeachType! and my missing type, I assume. Unfortunately, I only see the filename to my source file but didn't found out the actual line number of the call yet (The data I found hat a 0 line number).

On Monday, 14 February 2022 at 11:37:19 UTC, frame wrote:

The error may occurred as I re-organized my modules and a type became unavailable(?)

The typeSemantic function is called with a null-argument indeed:

type = null
loc.filename = "..\dmd2\src\phobos\std\array.d"

Something related to Appender.put!, to precise this line (3513):

auto bigData = (() @trusted => _data.arr.ptr[0 .. len + 1])();
               ^

I just ran into the same issue but this time I attached the debugger to a running process and step into some function. The error then occurred after loading a source file but I'm sure it has nothing to do with the source itself, it just happens from time to time. It dumps me a 2.5G file.

Forums