How to use the "native" backend? #1182
Comments
|
( possible duplicate) Do You see Rodney branch? And I sent some details by e-mail. |
|
Yes I saw your emails. No I haven't tried Rodney's branch. If there are good fixes on the Rodney branch, why aren't they integrated into the master? |
|
I think that best ask Rodney
…04.06.2024, 19:11, "Mika Nyström" ***@***.***>:
Yes I saw your emails. No I haven't tried Rodney's branch.
If there are good fixes on the Rodney branch, why aren't they integrated into the master?
|
|
I have long thought I had merged all my changes (first committed on branch rodney-new)
long ago. After Victor asked for them, I did a further check and it looked to me
like they were there in the head.
I have struggled greatly over merging with git in past. One realization
that has come to be more recently is that git's so-called "remotes" aren't remote,
they are local. But rather local copies of something that is remote, whose identity
is stored with the local copy, and there is some git command (I've forgotten its name) in
the local repo that will bring them\ "remotes" up-to-date with any changes made since to the
original. This realization has cleared up some amount of confusion on my part, but
it looks like I may be still confused about git.
I could well have made mistakes twice on this. Apparently Victor believes so.
As I remember, the process I had finally chosen was to do a rebase that brought
github head changes into my rodney-new branch on my computer. Then push rodney-new to
github, which triggers testing in rodney-new on github. Then if that's clean,
do something or other to move that into the head, quickly before more changes
came into the head on github.
Is "master" the right branch on github to try to get updated?
I would welcome any advice on what if anything is needed to get this resolved.
…On 6/4/24 11:11, Mika Nyström wrote:
Yes I saw your emails. No I haven't tried Rodney's branch.
If there are good fixes on the Rodney branch, why aren't they integrated into the master?
—
Reply to this email directly, view it on GitHub <#1182 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABSVZNFUIFBT4MZEYRZIU4LZFXRMNAVCNFSM6AAAAABIXRZSRWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBXHEYTKNJVGA>.
You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>
|
|
I have to admit git can confuse me, too, at times.
Here is what we see right now:
[image: RodneyAhead.png]
On Tue, Jun 4, 2024 at 1:31 PM Rodney M. Bates ***@***.***>
wrote:
… I have long thought I had merged all my changes (first committed on branch
rodney-new)
long ago. After Victor asked for them, I did a further check and it
looked to me
like they were there in the head.
I have struggled greatly over merging with git in past. One realization
that has come to be more recently is that git's so-called "remotes" aren't
remote,
they are local. But rather local copies of something that is remote,
whose identity
is stored with the local copy, and there is some git command (I've
forgotten its name) in
the local repo that will bring them\ "remotes" up-to-date with any changes
made since to the
original. This realization has cleared up some amount of confusion on my
part, but
it looks like I may be still confused about git.
I could well have made mistakes twice on this. Apparently Victor believes
so.
As I remember, the process I had finally chosen was to do a rebase that
brought
github head changes into my rodney-new branch on my computer. Then push
rodney-new to
github, which triggers testing in rodney-new on github. Then if that's
clean,
do something or other to move that into the head, quickly before more
changes
came into the head on github.
Is "master" the right branch on github to try to get updated?
I would welcome any advice on what if anything is needed to get this
resolved.
On 6/4/24 11:11, Mika Nyström wrote:
>
> Yes I saw your emails. No I haven't tried Rodney's branch.
>
> If there are good fixes on the Rodney branch, why aren't they integrated
into the master?
>
> —
> Reply to this email directly, view it on GitHub <
#1182 (comment)>, or
unsubscribe <
https://github.com/notifications/unsubscribe-auth/ABSVZNFUIFBT4MZEYRZIU4LZFXRMNAVCNFSM6AAAAABIXRZSRWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBXHEYTKNJVGA>.
> You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
>
—
Reply to this email directly, view it on GitHub
<#1182 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABKYNJLMABYX76WYZPY6KJTZFYPYXAVCNFSM6AAAAABIXRZSRWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBYGM3DMNJRGA>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Ah images not supported. Drat. For branch "rodney-new", github says This branch is 23 commits ahead of, 37 commits behind master. see |
|
( I answer by e-mail. I will fix formatting tomorrow)
((
} Is "master" the right branch on github to try to get updated?
))
Yes, master is right branch
…
04.06.2024, 23:31, "Rodney M. Bates" ***@***.***>:
I have long thought I had merged all my changes (first committed on branch rodney-new)
long ago. After Victor asked for them, I did a further check and it looked to me
like they were there in the head.
I have struggled greatly over merging with git in past. One realization
that has come to be more recently is that git's so-called "remotes" aren't remote,
they are local. But rather local copies of something that is remote, whose identity
is stored with the local copy, and there is some git command (I've forgotten its name) in
the local repo that will bring them\ "remotes" up-to-date with any changes made since to the
original. This realization has cleared up some amount of confusion on my part, but
it looks like I may be still confused about git.
I could well have made mistakes twice on this. Apparently Victor believes so.
As I remember, the process I had finally chosen was to do a rebase that brought
github head changes into my rodney-new branch on my computer. Then push rodney-new to
github, which triggers testing in rodney-new on github. Then if that's clean,
do something or other to move that into the head, quickly before more changes
came into the head on github.
Is "master" the right branch on github to try to get updated?
I would welcome any advice on what if anything is needed to get this resolved.
On 6/4/24 11:11, Mika Nyström wrote:
>
> Yes I saw your emails. No I haven't tried Rodney's branch.
>
> If there are good fixes on the Rodney branch, why aren't they integrated into the master?
>
> —
|
|
I have add images by Web interface of GitHub some time ago ( as small remark)
…
04.06.2024, 23:37, "Mika Nyström" ***@***.***>:
Ah images not supported. Drat.
For branch "rodney-new", github says
This branch is 23 commits ahead of, 37 commits behind master.
see
https://github.com/modula3/cm3/tree/rodney-new
|
|
As small remark: yes, yes we should synchronize _all_ useful branches as quick as possible.
…
04.06.2024, 23:36, "Mika Nyström" ***@***.***>:
I have to admit git can confuse me, too, at times.
Here is what we see right now:
[image: RodneyAhead.png]
On Tue, Jun 4, 2024 at 1:31 PM Rodney M. Bates ***@***.***>
wrote:
> I have long thought I had merged all my changes (first committed on branch
> rodney-new)
> long ago. After Victor asked for them, I did a further check and it
> looked to me
> like they were there in the head.
>
> I have struggled greatly over merging with git in past. One realization
> that has come to be more recently is that git's so-called "remotes" aren't
> remote,
> they are local. But rather local copies of something that is remote,
> whose identity
> is stored with the local copy, and there is some git command (I've
> forgotten its name) in
> the local repo that will bring them\ "remotes" up-to-date with any changes
> made since to the
> original. This realization has cleared up some amount of confusion on my
> part, but
> it looks like I may be still confused about git.
>
> I could well have made mistakes twice on this. Apparently Victor believes
> so.
>
> As I remember, the process I had finally chosen was to do a rebase that
> brought
> github head changes into my rodney-new branch on my computer. Then push
> rodney-new to
> github, which triggers testing in rodney-new on github. Then if that's
> clean,
> do something or other to move that into the head, quickly before more
> changes
> came into the head on github.
>
> Is "master" the right branch on github to try to get updated?
>
> I would welcome any advice on what if anything is needed to get this
> resolved.
>
> On 6/4/24 11:11, Mika Nyström wrote:
> >
> > Yes I saw your emails. No I haven't tried Rodney's branch.
> >
> > If there are good fixes on the Rodney branch, why aren't they integrated into the master?
> >
|
|
On 6/4/24 15:37, Mika Nyström wrote:
Ah images not supported. Drat.
For branch "rodney-new", github says
This branch is 23 commits ahead of <master...rodney-new>, 37 commits behind <rodney-new...master> master.
Ah, this -------^ at least shows me what is missing from master.
None of it sounds from the titles likely to affect the build process.
…
see
https://github.com/modula3/cm3/tree/rodney-new
—
Reply to this email directly, view it on GitHub <#1182 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABSVZNBC3ZJ4FF3GIWMUHUTZFYQRZAVCNFSM6AAAAABIXRZSRWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBYGM3TKOJRHE>.
You are receiving this because you commented.Message ID: ***@***.***>
|
|
We don't have "broken process" / {{ doing something step by step }}
We have broken results, i.e. compiled binaries
… 04.06.2024, 23:54, "Rodney M. Bates" ***@***.***>:
On 6/4/24 15:37, Mika Nyström wrote:
>
> Ah images not supported. Drat.
>
> For branch "rodney-new", github says
>
> This branch is 23 commits ahead of <master...rodney-new>, 37 commits behind <rodney-new...master> master.
>
Ah, this -------^ at least shows me what is missing from master.
None of it sounds from the titles likely to affect the build process.
>
> see
>
> https://github.com/modula3/cm3/tree/rodney-new
>
> —
|
|
Hi Rodney,
just a few notes on Git concepts as I think that may be helpful. If you
already know this,
just skip it.
Git uses a distributed version control architecture in which you have
local copies of
branches on remote sites. This is necessary to keep the naming of
branches unique.
When you clone a repository to work on it, by default you get a local
`remote` called
`origin`.
Local branches can be set to _track_ remote branches, which means
changes will automatically
be related in your local repository. This can either be done implicitly
by using
`git checkout <branch>` or explicitly by using `git branch
—set-upstream-to …`.
If you commit a change it will be committed to a local branch and your
upstream copies.
However, it will not be replicated to the original remote repository.
To move changes between your local and remote repositories use `git
fetch/pull/push`.
`fetch` will update one, several or all copies of remote branches to the
current remote
state. `push` will copy your remote branch to its upstream counterpart.
`pull` is actuallly
a combination of `fetch` and `checkout`, `merge` or `rebase`. It updates
both your local
counterparts of a remote branch and your actual local branch (without a
remote prefix).
`fetch` and `push` only work on the version store (in the .git
subdirectory), while pull
also updates your current workspace.
It’s a very flexible but also sometimes confusing design I admit.
So to get changes to the remote master or main branch, you have to do
something like
```
git checkout master
git merge <my branch>
git commit -a -m ‘…’
git push
```
I hope this helps,
Olaf
…On 4 Jun 2024, at 22:31, Rodney M. Bates wrote:
I have long thought I had merged all my changes (first committed on
branch rodney-new)
long ago. After Victor asked for them, I did a further check and it
looked to me
like they were there in the head.
I have struggled greatly over merging with git in past. One
realization
that has come to be more recently is that git's so-called "remotes"
aren't remote,
they are local. But rather local copies of something that is remote,
whose identity
is stored with the local copy, and there is some git command (I've
forgotten its name) in
the local repo that will bring them\ "remotes" up-to-date with any
changes made since to the
original. This realization has cleared up some amount of confusion
on my part, but
it looks like I may be still confused about git.
I could well have made mistakes twice on this. Apparently Victor
believes so.
As I remember, the process I had finally chosen was to do a rebase
that brought
github head changes into my rodney-new branch on my computer. Then
push rodney-new to
github, which triggers testing in rodney-new on github. Then if
that's clean,
do something or other to move that into the head, quickly before more
changes
came into the head on github.
Is "master" the right branch on github to try to get updated?
I would welcome any advice on what if anything is needed to get this
resolved.
On 6/4/24 11:11, Mika Nyström wrote:
>
> Yes I saw your emails. No I haven't tried Rodney's branch.
>
> If there are good fixes on the Rodney branch, why aren't they
> integrated into the master?
>
> —
> Reply to this email directly, view it on GitHub
> <#1182 (comment)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/ABSVZNFUIFBT4MZEYRZIU4LZFXRMNAVCNFSM6AAAAABIXRZSRWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNBXHEYTKNJVGA>.
> You are receiving this because you are subscribed to this
> thread.Message ID: ***@***.***>
>
--
Reply to this email directly or view it on GitHub:
#1182 (comment)
You are receiving this because you are subscribed to this thread.
Message ID: ***@***.***>
--
Olaf Wagner -- elego Software Solutions GmbH -- http://www.elegosoft.com
Gustav-Meyer-Allee 25 / Gebäude 12, 13355 Berlin,
Germany
phone: +49 30 23 45 86 96 mobile: +49 163 52 11 579 fax: +49 30 23 45
86 95
Geschäftsführer: Olaf Wagner | Sitz: Berlin
Handelregister: Amtsgericht Charlottenburg HRB 77719 | USt-IdNr:
DE163214194
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Does anyone out there know how we are supposed to use the "native" backend?
I am finding that my old code was not working with the C-based backend, because something goes wrong when I pass data from M3 to C.
I bootstrapped using an old CM3 and do-cm3-all.py
This seems to build everything using the C backend.
Then I built m3cc using the C-based backend.
This had the problem that when I try to build code using the native backend, it "sometimes" has issues linking against the CM3 libs generated by the C backend.
So it seems that the CM3 libs have to be rebuilt using the native backend if one wants to build one's own programs with the native backend.
Is this expected behavior?
To accomplish this, I had to change pylib.py:
diff --git i/scripts/python/pylib.py w/scripts/python/pylib.py
index e6e45000d..faa11b3a3 100755
--- i/scripts/python/pylib.py
+++ w/scripts/python/pylib.py
@@ -612,8 +612,8 @@ Target = Target or getenv("CM3_TARGET") or Host
_CBackend = "c" in LowercaseArgv or "+c" in LowercaseArgv or TargetOnlyHasCBackend(Target)
-if _CBackend:
+#if _CBackend:
+# CM3_FLAGS = CM3_FLAGS + " -DM3_BACKEND_MODE=C"
#-----------------------------------------------------------------------------
(Maybe there is a better way?)
And then I can rebuild all the CM3 libs using the native backend.
Now things actually work pretty well...!
Any comments from the mailing list?
I can try to make a small example of a lib built with the C backend that fails miserably when linking against native code.
I am seeing this ...
/scc406110> /nfs/site/home/mnystroe/char8/m3utils-3/mscheme/sstubgen/program/AMD64_LINUX/sstubgen
sstubgen: ../src/time/POSIX/DatePosixC.c:62: DatePosix__FromTime: Assertion `zone == Local || zone == UTC' failed.
Abort (core dumped)
Using host libthread_db library "/lib64/libthread_db.so.1".
sstubgen: ../src/time/POSIX/DatePosixC.c:62: DatePosix__FromTime: Assertion `zone == Local || zone == UTC' failed.
Program received signal SIGABRT, Aborted.
0x00002aaaab73a0c7 in raise () from /lib64/libc.so.6
Missing separate debuginfos, use: zypper install glibc-debuginfo-2.22-114.22.1.x86_64
(gdb) where
#0 0x00002aaaab73a0c7 in raise () from /lib64/libc.so.6
#1 0x00002aaaab73b49a in abort () from /lib64/libc.so.6
#2 0x00002aaaab733016 in __assert_fail_base () from /lib64/libc.so.6
#3 0x00002aaaab7330c2 in __assert_fail () from /lib64/libc.so.6
#4 0x000000000070e3c3 in DatePosix__FromTime (t=1717465520.698544, zone=140737488341680, date=0x7fffffffc9d0) at ../src/time/POSIX/DatePosixC.c:62
#5 0x000000000070dbf6 in Date__FromTime (_result_L_50=0x0, t_L_51=1717465520.698544, z_L_52=0x7fffffffca80 "\260\312\377\377\377\177") at ../src/time/POSIX/DatePosix.m3:29
#6 0x00000000004843a2 in FinDate__FromTime (M3_CtKayy_t=<error reading variable: inner value printing not implemented for language "Auto">,
M3_Ab1PrR_zone=<error reading variable: inner value printing not implemented for language "Auto">, M3_BGtbSH__result=<error reading variable: inner value printing not implemented for language "Auto">)
at ../src/FinDate.m3:190
#7 0x00000000004843f6 in FinDate__Today (M3_Ab1PrR_zone=<error reading variable: inner value printing not implemented for language "Auto">,
M3_BGtbSH__result=<error reading variable: inner value printing not implemented for language "Auto">) at ../src/FinDate.m3:186
#8 0x0000000000484d70 in FinDate_M3 (M3_AcxOUs_mode=<error reading variable: inner value printing not implemented for language "Auto">) at ../src/FinDate.m3:137
#9 0x00000000006e8620 in RTLinker__RunMainBody (m_L_92=0xa58680 <MI_FinDate>) at ../src/runtime/common/RTLinker.m3:409
#10 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa58a80 <MI_HMTime>) at ../src/runtime/common/RTLinker.m3:389
#11 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa58460 <MM_SafeTZ>) at ../src/runtime/common/RTLinker.m3:389
#12 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa57fc0 <MI_TZ>) at ../src/runtime/common/RTLinker.m3:389
#13 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa536c0 <MM_Debug>) at ../src/runtime/common/RTLinker.m3:389
#14 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa53580 <MI_Debug>) at ../src/runtime/common/RTLinker.m3:389
#15 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa4cd80 <MM_SchemeUtils>) at ../src/runtime/common/RTLinker.m3:389
#16 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa4cc00 <MI_SchemeUtils>) at ../src/runtime/common/RTLinker.m3:389
#17 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa4cac0 <MM_SchemePair>) at ../src/runtime/common/RTLinker.m3:389
#18 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa4c960 <MI_SchemePair>) at ../src/runtime/common/RTLinker.m3:389
#19 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa4c7a0 <MI_SchemeClass>) at ../src/runtime/common/RTLinker.m3:389
#20 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa4ba40 <MM_Scheme>) at ../src/runtime/common/RTLinker.m3:389
#21 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa4b7c0 <MI_Scheme>) at ../src/runtime/common/RTLinker.m3:389
#22 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa50d60 <MM_SchemeBoolean>) at ../src/runtime/common/RTLinker.m3:389
#23 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa50ca0 <MI_SchemeBoolean>) at ../src/runtime/common/RTLinker.m3:389
#24 0x00000000006e8435 in RTLinker__RunMainBody (m_L_92=0xa3e9c0 <MM_Main>) at ../src/runtime/common/RTLinker.m3:389
#25 0x00000000006e7316 in RTLinker__AddUnitI (m_L_95=0xa3e9c0 <MM_Main>) at ../src/runtime/common/RTLinker.m3:116
#26 0x00000000006e7433 in RTLinker__AddUnit (b_L_83=0x418f93 <Main_M3>) at ../src/runtime/common/RTLinker.m3:125
#27 0x00000000004183f8 in main (argc=1, argv=0x7fffffffd698, envp=0x7fffffffd6a8) at _m3main.cpp:53
(gdb)
The text was updated successfully, but these errors were encountered: