illegal recursive declaration crashes frontend

[demoitem]

This code from e019 CONST N = N + 1l;

causes a segv in the frontend from a stack overflow caused (ironically) by infinite recursion between Expr.RepTypeOf and AddExpr.RepTypeOf similar segv for N - 1, N * 1, N DIV 1 but strangely not N MOD 1

[RodneyBates]

Hmm, This works OK for me, using the compiler in branch rodney-new. I was under the belief that everything in rodney-new was merged and pushed into master on github.  I'll look into getting things merged.

By "works", I mean produces the compile error, but no segfault.

On 10/17/22 21:05, peter mckinna wrote:

This code from e019 CONST N = N + 1l;

causes a segv in the frontend from a stack overflow caused (ironically) by infinite recursion between Expr.RepTypeOf and AddExpr.RepTypeOf similar segv for N - 1, N * 1, N DIV 1 but strangely not N MOD 1

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSVZNF6OOTS2NYHLPDBWVDWDYAVLANCNFSM6AAAAAARHTSKR4. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[demoitem]

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

On Wed, 19 Oct 2022 at 10:07, Rodney M. Bates @.***> wrote:

Hmm, This works OK for me, using the compiler in branch rodney-new. I was under the belief that everything in rodney-new was merged and pushed into master on github. I'll look into getting things merged.

By "works", I mean produces the compile error, but no segfault.

On 10/17/22 21:05, peter mckinna wrote:

This code from e019 CONST N = N + 1l;

causes a segv in the frontend from a stack overflow caused (ironically) by infinite recursion between Expr.RepTypeOf and AddExpr.RepTypeOf similar segv for N - 1, N * 1, N DIV 1 but strangely not N MOD 1

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABSVZNF6OOTS2NYHLPDBWVDWDYAVLANCNFSM6AAAAAARHTSKR4 . You are receiving this because you are subscribed to this thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1283110996, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABALZX5YE3UUDLLVKYBHSWLWD4UTTANCNFSM6AAAAAARHTSKR4 . You are receiving this because you authored the thread.Message ID: @.***>

[VictorMiasnikov]

( . . .) This works OK for me, using the compiler in branch rodney-new. I was under the belief that everything in rodney-new was merged and pushed into master on github.  I'll look into getting things merged. By "works", I mean produces the compile error, but no segfault.

In short: please do it ( merge)

Thank!

[VictorMiasnikov]

(

a-la P.S.

No worries

Why?

"branch rodney-new" tested as part of 2-3 latest releases with good results.

https://github.com/VictorMiasnikov/cm3/commit/a5580bdce9ebd91c57b9e13314d8337f3563b9a0 =} https://github.com/VictorMiasnikov/cm3/actions/runs/3274227685

)

[VictorMiasnikov]

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

I.e. before "merge branch rodney-new" we have:

https://github.com/modula3/cm3/commit/844d77c22fe7845d0e0b0df4f8ebcc857594627a ( https://github.com/modula3/cm3/commit/844d77c22fe7845d0e0b0df4f8ebcc857594627a )

https://github.com/modula3/cm3/actions/runs/3231157773 2022-10-12T01:06:46.2343570Z

--- e020 --- illegal recursive declaration
cd ../src/e0/e020 && cm3 -silent -DM3TESTS  >AMD64_DARWIN/stdout.build.raw 2>AMD64_DARWIN/stderr.build.raw
--- ../src/e0/e020/stdout.build 2022-10-12 00:36:07.000000000 +0000
+++ ../src/e0/e020/AMD64_DARWIN/stdout.build    2022-10-12 01:06:46.000000000 +0000
@@ -1 +0,0 @@
-Fatal Error: package build failed
--- ../src/e0/e020/stderr.build 2022-10-12 00:36:07.000000000 +0000
+++ ../src/e0/e020/AMD64_DARWIN/stderr.build    2022-10-12 01:06:46.000000000 +0000
@@ -0,0 +1 @@
+sh: line 1: 31473 Segmentation fault: 11  ./pgm > stdout.pgm.raw 2> stderr.pgm.raw

But after integration we have:

2022-10-18T14:42:54.2926734Z

--- e020 --- illegal recursive declaration
cd ../src/e0/e020 && cm3 -silent -DM3TESTS  >AMD64_LINUX/stdout.build.raw 2>AMD64_LINUX/stderr.build.raw

[VictorMiasnikov]

This code from e019 CONST N = N + 1l;

causes a segv in the frontend from a stack overflow caused (ironically) by infinite recursion between Expr.RepTypeOf and AddExpr.RepTypeOf similar segv for N - 1, N * 1, N DIV 1 but strangely not N MOD 1

I.e. before:

2022-10-12T01:06:47.0713330Z
--- e019 --- illegal recursive declaration
cd ../src/e0/e019 && cm3 -silent -DM3TESTS  >AMD64_DARWIN/stdout.build.raw 2>AMD64_DARWIN/stderr.build.raw
sh: line 1: 31493 Segmentation fault: 11  cm3 -silent -DM3TESTS > AMD64_DARWIN/stdout.build.raw 2> AMD64_DARWIN/stderr.build.raw
--- ../src/e0/e019/stdout.build 2022-10-12 00:36:07.000000000 +0000
+++ ../src/e0/e019/AMD64_DARWIN/stdout.build    2022-10-12 01:06:47.000000000 +0000
@@ -1,5 +1 @@
 "../Main.m3", line 9: illegal recursive declaration (N)
-"../Main.m3", line 9: unsafe '+'
-"../Main.m3", line 9: value is not constant
-3 errors encountered
-Fatal Error: package build failed

After:

cd ../src/e0/e019 && cm3 -silent -DM3TESTS  >AMD64_LINUX/stdout.build.raw 2>AMD64_LINUX/stderr.build.raw
--- ../src/e0/e019/stdout.build 2022-10-18 14:34:14.867453360 +0000
+++ ../src/e0/e019/AMD64_LINUX/stdout.build 2022-10-18 14:42:55.101618323 +0000
@@ -1,3 +1 @@
-"../Main.m3", line 9: Illegal recursive declaration (2.4.8). (N)
-1 error encountered
-Fatal Error: package build failed
+"../Main.m3", line 9: illegal recursive declaration (N)
--- ../src/e0/e019/stderr.build 2022-10-18 14:34:14.867453360 +0000
+++ ../src/e0/e019/AMD64_LINUX/stderr.build 2022-10-18 14:42:55.101618323 +0000

P.S.

This looks like changing "error message" text:

-"../Main.m3", line 9: Illegal recursive declaration (2.4.8). (N)
+"../Main.m3", line 9: illegal recursive declaration (N)

[VictorMiasnikov]

P.P.S.

https://github.com/VictorMiasnikov/cm3/actions/runs/3279987990/jobs/5400217924

I386_NT  build (integrated)

--- e019 --- illegal recursive declaration
cd ../src/e0/e019 && cm3 -silent -DM3TESTS  >I386_NT/stdout.build.raw 2>I386_NT/stderr.build.raw
Comparing files ..\SRC\E0\E019\stdout.build and ..\SRC\E0\E019\I386_NT\STDOUT.BUILD
***** ..\SRC\E0\E019\stdout.build
"../Main.m3", line 9: Illegal recursive declaration (2.4.8). (N)
1 error encountered
Fatal Error: package build failed
***** ..\SRC\E0\E019\I386_NT\STDOUT.BUILD
"../Main.m3", line 9: illegal recursive declaration (2.4.8). (N)
*****

Comparing files ..\SRC\E0\E019\stderr.build and ..\SRC\E0\E019\I386_NT\STDERR.BUILD
***** ..\SRC\E0\E019\stderr.build
m3front failed compiling: ../Main.m3
***** ..\SRC\E0\E019\I386_NT\STDERR.BUILD

***
*** runtime error:
***    A runtime error occurred.
***    pc = 0x120b700 = Get + 0x9 in ../src/runtime/common/RTType.m3
***

Stack trace:
   FP         PC      Procedure
---------  ---------  -------------------------------
0x1e02fe8  0x1227cc4  SystemError + 0x64 in ../src/runtime/WIN32/RTSignal.m3
0x1e0300c  0x120b700  Get + 0x9 in ../src/runtime/common/RTType.m3
0x1e0302c  0x120b57a  IsSubtype + 0x2e in ../src/runtime/common/RTType.m3
0x1e03068  0x120ba48  ScanTypecase + 0x12c in ../src/runtime/common/RTType.m3
0x1e03094  0x110f323  Split + 0x20 in ../src/exprs/NamedExpr.m3
0x1e030c8  0x10f4e35  StripNamedCons + 0x33 in ../src/exprs/Expr.m3
0x1e030f8  0x10f303d  RepTypeOf + 0x1c in ../src/exprs/Expr.m3
0x1e03124  0x114fec0  RepTypeOf + 0x5d in ../src/exprs/AddExpr.m3
0x1e03154  0x10f30e1  RepTypeOf + 0xc0 in ../src/exprs/Expr.m3
0x1e03180  0x114fec0  RepTypeOf + 0x5d in ../src/exprs/AddExpr.m3
.........  .........  ... more frames ...
*****

The system cannot find the file specified.

+

https://github.com/VictorMiasnikov/cm3/actions/runs/3280163012/jobs/5400877203

AMD64_LINUX gcc

--- ../src/e0/e019/stdout.build 2022-10-19 08:59:44.623386187 +0000
+++ ../src/e0/e019/AMD64_LINUX/stdout.build 2022-10-19 09:14:29.163145789 +0000
@@ -1,3 +1 @@
 "../Main.m3", line 9: Illegal recursive declaration (2.4.8). (N)
-1 error encountered
-Fatal Error: package build failed
--- ../src/e0/e019/stderr.build 2022-10-19 08:59:44.623386187 +0000
+++ ../src/e0/e019/AMD64_LINUX/stderr.build 2022-10-19 09:14:29.163145789 +0000
@@ -1 +1 @@
-m3front failed compiling: ../Main.m3
+Segmentation fault

P.P.P.S.

"Combo branch":

https://github.com/VictorMiasnikov/cm3/releases/tag/d5.11.9-ZZYYXX-20221019_15-00

[RodneyBates]

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+  page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch.  I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric    https://github.com/coder5506/cm3.git (fetch) eric    https://github.com/coder5506/cm3.git (push) origin    /home/rodney/proj/m3/git-devel/cm3 (fetch) origin    /home/rodney/proj/m3/git-devel/cm3 (push) upstream    https://github.com/modula3/cm3.git (fetch) upstream    https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric    https://github.com/coder5506/cm3.git (fetch) eric    https://github.com/coder5506/cm3.git (push) origin    /home/rodney/proj/m3/git-devel/cm3 (fetch) origin    /home/rodney/proj/m3/git-devel/cm3 (push) upstream    https://github.com/modula3/cm3.git (fetch) upstream    https://github.com/modula3/cm3.git (push) @.**:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3    57ee5febdb..65ca3e1248  master     -> upstream/master   [new branch]            zcxh       -> upstream/zcxh

apparently successfully.  so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both.  Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

[RodneyBates]

PS:

1. I am quite sure this is the way I did this last time, some months ago.

2. I have definitely made no changes to my local master copy since then.

3. I pushed my rodney-new branch up to github, in case somebody has an    urgent need for its contents.

4. I added a bunch of cases to e019.  All give compile errors, as expected    using my latest compiler, built from rodney-new, but don't crash.  A    couple give a second, probably extraneous, compile message.

On 10/24/22 15:01, Rodney Bates wrote:

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+  page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch.  I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric    https://github.com/coder5506/cm3.git (fetch) eric    https://github.com/coder5506/cm3.git (push) origin    /home/rodney/proj/m3/git-devel/cm3 (fetch) origin    /home/rodney/proj/m3/git-devel/cm3 (push) upstream    https://github.com/modula3/cm3.git (fetch) upstream    https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric    https://github.com/coder5506/cm3.git (fetch) eric    https://github.com/coder5506/cm3.git (push) origin    /home/rodney/proj/m3/git-devel/cm3 (fetch) origin    /home/rodney/proj/m3/git-devel/cm3 (push) upstream    https://github.com/modula3/cm3.git (fetch) upstream    https://github.com/modula3/cm3.git (push) @.**:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3    57ee5febdb..65ca3e1248  master     -> upstream/master   [new branch]            zcxh       -> upstream/zcxh

apparently successfully.  so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both.  Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

[demoitem]

Eric may have better solutions but I do the following.

If I have been making minor changes on master make sure you are on the master branch on your local machine.

git checkout master git branch (to make sure) git fetch git pull --rebase origin master (rebase creates commits on your local master so your changes are the latest - may provoke conflicts) git push origin master (sends your local master to remote)

If I have created a local branch I merge on my local machine first then do the push.

git checkout master (make sure you are on master locally) git fetch git pull --rebase origin master git merge other-branch (may have conflicts gti tells you what to do) git commit -a ( creates a merge commit) git push origin master (or just git push works for me)

now master branch should be the same on local and remote.

git log (verify your commits are in place)

On Tue, 25 Oct 2022 at 07:30, Rodney M. Bates @.***> wrote:

PS:

1. I am quite sure this is the way I did this last time, some months ago.

2. I have definitely made no changes to my local master copy since then.

3. I pushed my rodney-new branch up to github, in case somebody has an urgent need for its contents.

4. I added a bunch of cases to e019. All give compile errors, as expected using my latest compiler, built from rodney-new, but don't crash. A couple give a second, probably extraneous, compile message.

On 10/24/22 15:01, Rodney Bates wrote:

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+ page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch. I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push) @.***:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3 57ee5febdb..65ca3e1248 master -> upstream/master

• [new branch] zcxh -> upstream/zcxh

apparently successfully. so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both. Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289576520, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABALZX34KOPHPTMBMK22FETWE3WXVANCNFSM6AAAAAARHTSKR4 . You are receiving this because you authored the thread.Message ID: @.***>

[RodneyBates]

I have been using an even more conservative (I think) procedure.  I alter local master only from github master.  Then merge or rebase into my local branch rodney-new, push that up to github rodney-new, and let the CI tests run on it.  If they are good enough, then I push local rodney-new to github master.  But now I can't even do step one.

On 10/24/22 16:05, peter mckinna wrote:

Eric may have better solutions but I do the following.

If I have been making minor changes on master make sure you are on the master branch on your local machine.

git checkout master git branch (to make sure) git fetch git pull --rebase origin master (rebase creates commits on your local master so your changes are the latest - may provoke conflicts) git push origin master (sends your local master to remote)

If I have created a local branch I merge on my local machine first then do the push.

git checkout master (make sure you are on master locally) git fetch git pull --rebase origin master git merge other-branch (may have conflicts gti tells you what to do) git commit -a ( creates a merge commit) git push origin master (or just git push works for me)

now master branch should be the same on local and remote.

git log (verify your commits are in place)

On Tue, 25 Oct 2022 at 07:30, Rodney M. Bates @.***> wrote:

PS:

1. I am quite sure this is the way I did this last time, some months ago.

2. I have definitely made no changes to my local master copy since then.

3. I pushed my rodney-new branch up to github, in case somebody has an urgent need for its contents.

4. I added a bunch of cases to e019. All give compile errors, as expected using my latest compiler, built from rodney-new, but don't crash. A couple give a second, probably extraneous, compile message.

On 10/24/22 15:01, Rodney Bates wrote:

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+ page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch. I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push) @.***:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3 57ee5febdb..65ca3e1248 master -> upstream/master

• [new branch] zcxh -> upstream/zcxh

apparently successfully. so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both. Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289576520, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABALZX34KOPHPTMBMK22FETWE3WXVANCNFSM6AAAAAARHTSKR4 . You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289615617, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSVZNDIIQRY5SENVUIWEODWE32ZFANCNFSM6AAAAAARHTSKR4. You are receiving this because you commented.Message ID: @.***>

[demoitem]

Your local rodney-new should be in sync with the remote, so just merge into master locally, then push. Are the CI testers making changes to rodney-new? I'm not sure about remote merging.

This whole CI thing might need examination. I get they are testing on all architectures but as far as I can tell they are only using the C backend whereas they should be testing other backends as well. Also which branches do they test? Surely not all branches. How would one get them to test a particular branch? And while we are at it, I think a bit of judicious pruning of some of those github branches would be beneficial.

On Tue, 25 Oct 2022 at 08:53, Rodney M. Bates @.***> wrote:

I have been using an even more conservative (I think) procedure. I alter local master only from github master. Then merge or rebase into my local branch rodney-new, push that up to github rodney-new, and let the CI tests run on it. If they are good enough, then I push local rodney-new to github master. But now I can't even do step one.

On 10/24/22 16:05, peter mckinna wrote:

Eric may have better solutions but I do the following.

If I have been making minor changes on master make sure you are on the master branch on your local machine.

git checkout master git branch (to make sure) git fetch git pull --rebase origin master (rebase creates commits on your local master so your changes are the latest - may provoke conflicts) git push origin master (sends your local master to remote)

If I have created a local branch I merge on my local machine first then do the push.

git checkout master (make sure you are on master locally) git fetch git pull --rebase origin master git merge other-branch (may have conflicts gti tells you what to do) git commit -a ( creates a merge commit) git push origin master (or just git push works for me)

now master branch should be the same on local and remote.

git log (verify your commits are in place)

On Tue, 25 Oct 2022 at 07:30, Rodney M. Bates @.***> wrote:

PS:

1. I am quite sure this is the way I did this last time, some months ago.

2. I have definitely made no changes to my local master copy since then.

3. I pushed my rodney-new branch up to github, in case somebody has an urgent need for its contents.

4. I added a bunch of cases to e019. All give compile errors, as expected using my latest compiler, built from rodney-new, but don't crash. A couple give a second, probably extraneous, compile message.

On 10/24/22 15:01, Rodney Bates wrote:

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+ page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch. I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push) @.***:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3 57ee5febdb..65ca3e1248 master -> upstream/master

• [new branch] zcxh -> upstream/zcxh

apparently successfully. so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both. Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289576520, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABALZX34KOPHPTMBMK22FETWE3WXVANCNFSM6AAAAAARHTSKR4

. You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289615617>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABSVZNDIIQRY5SENVUIWEODWE32ZFANCNFSM6AAAAAARHTSKR4 . You are receiving this because you commented.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289664620, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABALZX4HSACTPBLNRB6X2XTWE4AMZANCNFSM6AAAAAARHTSKR4 . You are receiving this because you authored the thread.Message ID: @.***>

[VictorMiasnikov]

Your local rodney-new should be in sync with the remote, so just merge into master locally, then push. Are the CI testers making changes to rodney-new? I'm not sure about remote merging.

As I known, only Rodney making changes in "rodney-new branch"

(

This whole CI thing might need examination. I get they are testing on all architectures but as far as I can tell they are only using the C backend whereas they should be testing other backends as well. Also which branches do they test? Surely not all branches. How would one get them to test a particular branch? And while we are at it, I think a bit of judicious pruning of some of those github branches would be beneficial.

Please re-post Your msg to issue N1102

)

[RodneyBates]

On 10/24/22 17:46, peter mckinna wrote:

Your local rodney-new should be in sync with the remote, so just merge into master locally, then push. Are the CI testers making changes to rodney-new?

I don't think so.

I prefer creating a merged (or rebased) local rodney-new and pushing it first, to avoid risking introducing breakage in github master. But.. regardless of that, before I can create a merge either way, I have to get github master to pull down, which is where I am stuck right now.  This used to work.

I'm not sure about remote merging.

This whole CI thing might need examination. I get they are testing on all architectures but as far as I can tell they are only using the C backend whereas they should be testing other backends as well. Also which branches do they test? \ As of the last time I pushed rodney-new, it provoked tests.

Surely not all branches. How would one get them to test a particular branch? And while we are at it, I think a bit of judicious pruning of some of those github branches would be beneficial.

On Tue, 25 Oct 2022 at 08:53, Rodney M. Bates @.***> wrote:

I have been using an even more conservative (I think) procedure. I alter local master only from github master. Then merge or rebase into my local branch rodney-new, push that up to github rodney-new, and let the CI tests run on it. If they are good enough, then I push local rodney-new to github master. But now I can't even do step one.

On 10/24/22 16:05, peter mckinna wrote:

Eric may have better solutions but I do the following.

If I have been making minor changes on master make sure you are on the master branch on your local machine.

git checkout master git branch (to make sure) git fetch git pull --rebase origin master (rebase creates commits on your local master so your changes are the latest - may provoke conflicts) git push origin master (sends your local master to remote)

If I have created a local branch I merge on my local machine first then do the push.

git checkout master (make sure you are on master locally) git fetch git pull --rebase origin master git merge other-branch (may have conflicts gti tells you what to do) git commit -a ( creates a merge commit) git push origin master (or just git push works for me)

now master branch should be the same on local and remote.

git log (verify your commits are in place)

On Tue, 25 Oct 2022 at 07:30, Rodney M. Bates @.***> wrote:

PS:

1. I am quite sure this is the way I did this last time, some months ago.

2. I have definitely made no changes to my local master copy since then.

3. I pushed my rodney-new branch up to github, in case somebody has an urgent need for its contents.

4. I added a bunch of cases to e019. All give compile errors, as expected using my latest compiler, built from rodney-new, but don't crash. A couple give a second, probably extraneous, compile message.

On 10/24/22 15:01, Rodney Bates wrote:

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+ page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch. I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push) @.***:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3 57ee5febdb..65ca3e1248 master -> upstream/master

• [new branch] zcxh -> upstream/zcxh

apparently successfully. so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both. Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289576520, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABALZX34KOPHPTMBMK22FETWE3WXVANCNFSM6AAAAAARHTSKR4

. You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289615617>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABSVZNDIIQRY5SENVUIWEODWE32ZFANCNFSM6AAAAAARHTSKR4 . You are receiving this because you commented.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289664620, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABALZX4HSACTPBLNRB6X2XTWE4AMZANCNFSM6AAAAAARHTSKR4 . You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289727310, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSVZNFFRN2YUCATIPIH2CDWE4GTZANCNFSM6AAAAAARHTSKR4. You are receiving this because you commented.Message ID: @.***>

[demoitem]

I didn't see a git pull in your earlier email. You did the fetch which just gets the commits but not the pull which puts them into the branch. Was that an oversight?

On Thu, 27 Oct 2022 at 03:35, Rodney M. Bates @.***> wrote:

On 10/24/22 17:46, peter mckinna wrote:

Your local rodney-new should be in sync with the remote, so just merge into master locally, then push. Are the CI testers making changes to rodney-new?

I don't think so.

I prefer creating a merged (or rebased) local rodney-new and pushing it first, to avoid risking introducing breakage in github master. But.. regardless of that, before I can create a merge either way, I have to get github master to pull down, which is where I am stuck right now. This used to work.

I'm not sure about remote merging.

This whole CI thing might need examination. I get they are testing on all architectures but as far as I can tell they are only using the C backend whereas they should be testing other backends as well. Also which branches do they test? \ As of the last time I pushed rodney-new, it provoked tests.

Surely not all branches. How would one get them to test a particular branch? And while we are at it, I think a bit of judicious pruning of some of those github branches would be beneficial.

On Tue, 25 Oct 2022 at 08:53, Rodney M. Bates @.***> wrote:

I have been using an even more conservative (I think) procedure. I alter local master only from github master. Then merge or rebase into my local branch rodney-new, push that up to github rodney-new, and let the CI tests run on it. If they are good enough, then I push local rodney-new to github master. But now I can't even do step one.

On 10/24/22 16:05, peter mckinna wrote:

Eric may have better solutions but I do the following.

If I have been making minor changes on master make sure you are on the master branch on your local machine.

git checkout master git branch (to make sure) git fetch git pull --rebase origin master (rebase creates commits on your local master so your changes are the latest - may provoke conflicts) git push origin master (sends your local master to remote)

If I have created a local branch I merge on my local machine first then do the push.

git checkout master (make sure you are on master locally) git fetch git pull --rebase origin master git merge other-branch (may have conflicts gti tells you what to do) git commit -a ( creates a merge commit) git push origin master (or just git push works for me)

now master branch should be the same on local and remote.

git log (verify your commits are in place)

On Tue, 25 Oct 2022 at 07:30, Rodney M. Bates @.***> wrote:

PS:

1. I am quite sure this is the way I did this last time, some months ago.

2. I have definitely made no changes to my local master copy since then.

3. I pushed my rodney-new branch up to github, in case somebody has an urgent need for its contents.

4. I added a bunch of cases to e019. All give compile errors, as expected using my latest compiler, built from rodney-new, but don't crash. A couple give a second, probably extraneous, compile message.

On 10/24/22 15:01, Rodney Bates wrote:

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+ page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch. I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push) @.***:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3 57ee5febdb..65ca3e1248 master -> upstream/master

• [new branch] zcxh -> upstream/zcxh

apparently successfully. so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both. Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289576520>, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ABALZX34KOPHPTMBMK22FETWE3WXVANCNFSM6AAAAAARHTSKR4

. You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289615617>, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ABSVZNDIIQRY5SENVUIWEODWE32ZFANCNFSM6AAAAAARHTSKR4

. You are receiving this because you commented.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289664620, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABALZX4HSACTPBLNRB6X2XTWE4AMZANCNFSM6AAAAAARHTSKR4

. You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289727310>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABSVZNFFRN2YUCATIPIH2CDWE4GTZANCNFSM6AAAAAARHTSKR4 . You are receiving this because you commented.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1292311346, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABALZX7ZJUSZN5HCHHXBDGTWFFMTVANCNFSM6AAAAAARHTSKR4 . You are receiving this because you authored the thread.Message ID: @.***>

[RodneyBates]

On 10/26/22 16:46, peter mckinna wrote:

I didn't see a git pull in your earlier email. You did the fetch which just gets the commits but not the pull which puts them into the branch. Was that an oversight?

No, no oversight.  Per Eric's advice from months ago, for getting my local master updated, I fetch, then rebase.  But in backscrolling, I see a mistake.  I actually did a merge upstream/master instead of a rebase, and that was what refused:

  :~/proj/m3/git-merge/cm3/master$ git merge upstream/master   fatal: Not possible to fast-forward, aborting.

Moreover, I remember doing some kind of config on merge, that makes it do fast-forward only, which explains the refusal to do it another way.

So I tried again with the rebase.  That refused too, but unlike merge, it told me about one simple change to my local master that explains why fast-forward was impossible. Fixing that, the rebase worked as expected.

So thanks for the question, as it made me look in the right place to discover the problem.

On Thu, 27 Oct 2022 at 03:35, Rodney M. Bates @.***> wrote:

On 10/24/22 17:46, peter mckinna wrote:

Your local rodney-new should be in sync with the remote, so just merge into master locally, then push. Are the CI testers making changes to rodney-new?

I don't think so.

I prefer creating a merged (or rebased) local rodney-new and pushing it first, to avoid risking introducing breakage in github master. But.. regardless of that, before I can create a merge either way, I have to get github master to pull down, which is where I am stuck right now. This used to work.

I'm not sure about remote merging.

This whole CI thing might need examination. I get they are testing on all architectures but as far as I can tell they are only using the C backend whereas they should be testing other backends as well. Also which branches do they test? \ As of the last time I pushed rodney-new, it provoked tests.

Surely not all branches. How would one get them to test a particular branch? And while we are at it, I think a bit of judicious pruning of some of those github branches would be beneficial.

On Tue, 25 Oct 2022 at 08:53, Rodney M. Bates @.***> wrote:

I have been using an even more conservative (I think) procedure. I alter local master only from github master. Then merge or rebase into my local branch rodney-new, push that up to github rodney-new, and let the CI tests run on it. If they are good enough, then I push local rodney-new to github master. But now I can't even do step one.

On 10/24/22 16:05, peter mckinna wrote:

Eric may have better solutions but I do the following.

If I have been making minor changes on master make sure you are on the master branch on your local machine.

git checkout master git branch (to make sure) git fetch git pull --rebase origin master (rebase creates commits on your local master so your changes are the latest - may provoke conflicts) git push origin master (sends your local master to remote)

If I have created a local branch I merge on my local machine first then do the push.

git checkout master (make sure you are on master locally) git fetch git pull --rebase origin master git merge other-branch (may have conflicts gti tells you what to do) git commit -a ( creates a merge commit) git push origin master (or just git push works for me)

now master branch should be the same on local and remote.

git log (verify your commits are in place)

On Tue, 25 Oct 2022 at 07:30, Rodney M. Bates @.***> wrote:

PS:

1. I am quite sure this is the way I did this last time, some months ago.

2. I have definitely made no changes to my local master copy since then.

3. I pushed my rodney-new branch up to github, in case somebody has an urgent need for its contents.

4. I added a bunch of cases to e019. All give compile errors, as expected using my latest compiler, built from rodney-new, but don't crash. A couple give a second, probably extraneous, compile message.

On 10/24/22 15:01, Rodney Bates wrote:

OK, excuse my frustration, but once again, I can't seem to get the most basic things done in git, despite a 300+ page book, a 400+ page book, and lots of helpful emails from last fall.

All I am trying at this point is to get my local master branch up to date with the github master branch. I have:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push)

and have done git fetch:

~/proj/m3/git-merge/cm3/master$ git remote -v eric https://github.com/coder5506/cm3.git (fetch) eric https://github.com/coder5506/cm3.git (push) origin /home/rodney/proj/m3/git-devel/cm3 (fetch) origin /home/rodney/proj/m3/git-devel/cm3 (push) upstream https://github.com/modula3/cm3.git (fetch) upstream https://github.com/modula3/cm3.git (push) @.***:~/proj/m3/git-merge/cm3/master$ git fetch upstream remote: Enumerating objects: 615, done. remote: Counting objects: 100% (614/614), done. remote: Compressing objects: 100% (231/231), done. remote: Total 615 (delta 373), reused 592 (delta 367), pack-reused 1 Receiving objects: 100% (615/615), 556.76 KiB | 1.66 MiB/s, done. Resolving deltas: 100% (373/373), completed with 27 local objects. From https://github.com/modula3/cm3 57ee5febdb..65ca3e1248 master -> upstream/master

• [new branch] zcxh -> upstream/zcxh

apparently successfully. so all the files of master on github should be copied locally inside my .git, and my "remote" (confusingly misnamed) named upstream should point to my local copy of the latest github.master.head.

But:

~/proj/m3/git-merge/cm3/master$ git merge upstream/master fatal: Not possible to fast-forward, aborting. @.***:~/proj/m3/git-merge/cm3/master$ git merge upstream master merge: upstream - not something we can merge

Previously, I remember discovering that some git commands call for "upstream/master" and others "upstream master" to name a (local copy of) a branch at a remote location, but I don't find any reference material telling me which, particularly for git merge, so I tried both. Looks like the slash was the right one, but why does it refuse and what to do?

On 10/18/22 21:17, peter mckinna wrote:

No worries,

I have had merge problems of my own in the past. While you are at it, have a look at e020. It compiles but crashes. The stderr.pgm file suggests it should produce a runtime error - invalid cycle in super types.

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289576520>, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ABALZX34KOPHPTMBMK22FETWE3WXVANCNFSM6AAAAAARHTSKR4

. You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289615617>, or unsubscribe <

https://github.com/notifications/unsubscribe-auth/ABSVZNDIIQRY5SENVUIWEODWE32ZFANCNFSM6AAAAAARHTSKR4

. You are receiving this because you commented.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1289664620, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABALZX4HSACTPBLNRB6X2XTWE4AMZANCNFSM6AAAAAARHTSKR4

. You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub < https://github.com/modula3/cm3/issues/1109#issuecomment-1289727310>, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABSVZNFFRN2YUCATIPIH2CDWE4GTZANCNFSM6AAAAAARHTSKR4 . You are receiving this because you commented.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1292311346, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABALZX7ZJUSZN5HCHHXBDGTWFFMTVANCNFSM6AAAAAARHTSKR4 . You are receiving this because you authored the thread.Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/modula3/cm3/issues/1109#issuecomment-1292689196, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSVZNHO42WKQZRYYEJ2KRDWFGRDDANCNFSM6AAAAAARHTSKR4. You are receiving this because you commented.Message ID: @.***>

[VictorMiasnikov]

It looks like, "it's time" to integrate rodney-new branch to master branch

Thank! Best regards , Victor Miasnikov