On 7/3/19 5:49 PM, James E. King III via Boost wrote:

On Wed, Jul 3, 2019 at 9:15 AM Paul A Bristow via Boost wrote:

...

I have been bitten by this problem, again ☹

error: Cannot create link boost\pending\detail\disjoint_sets.hpp to libs\graph\include\boost\pending\detail\disjoint_sets.hpp. error: Link previously defined to another file, libs\disjoint_sets\include\boost\pending\detail\disjoint_sets.hpp.

that was discussed

https://lists.boost.org/Archives/boost/2019/04/246083.php

I have got going again by deleting the contents of boost/boost/ containing the symlinks.

But I note that a folder pending still exists, but not graph.

Is this unresolved?

Should it be fixed for the next release?

Paul

Remove your local libs/disjoint_sets directory, that should resolve it.

Doesn't the disjoint_sets git submodule need to be removed from the superproject?

-----Original Message----- From: Boost On Behalf Of James E. King III via Boost Sent: 3 July 2019 16:48 To: Boost Developers List Cc: James E. King III Subject: Re: [boost] Link problem with graph and pending still appears unresolved? On Wed, Jul 3, 2019 at 11:26 AM Andrey Semashev via Boost wrote:

Doesn't the disjoint_sets git submodule need to be removed from the superproject?

It was removed from develop in https://github.com/boostorg/boost/pull/283

I naively did what I normally do to update to current develop branch git checkout develop git pull --recurse-submodules git submodule update --init followed by a boostrap.bat and b2 -a headers but perhaps that doesn't prune dead branches? Should I have done something else? (Apart from deleting all the symlinks - some of which are my own private ones and I want to keep so a bit more tricky. Paul

-----Original Message----- From: Boost On Behalf Of Peter Dimov via Boost Sent: 3 July 2019 17:49 To: boost@lists.boost.org Cc: Peter Dimov Subject: Re: [boost] Link problem with graph and pending still appears unresolved? Paul A Bristow wrote:

I naively did what I normally do to update to current develop branch

git checkout develop git pull --recurse-submodules git submodule update --init

followed by a boostrap.bat

and b2 -a headers

but perhaps that doesn't prune dead branches?

`git submodule update --init` creates directories as needed for new modules, but it doesn't delete directories as needed (for removed modules.) OK - it's a feature. We don't do deleting often, so it won't bite many people, and Google should help them find out what to do from this conversation. Thanks Paul

On 4/07/2019 07:35, Andrey Semashev wrote:

On 7/3/19 6:48 PM, James E. King III wrote:

...

On Wed, Jul 3, 2019 at 11:26 AM Andrey Semashev wrote:

...

Doesn't the disjoint_sets git submodule need to be removed from the superproject?

It was removed from develop in https://github.com/boostorg/boost/pull/283

Strangely enough, it is still present in my .git/config, even after `git pull` and `git submodule update` in the Boost root.

What exists in your config file is as a result of the submodule init commands, and does not get cleared because upstream removed the submodule. The thing to check whether upstream still wants the submodule to exist or not is the contents of the .gitmodules file. Once you have confirmed that the module is no longer listed in .gitmodules, run: git submodule deinit module-name to remove it from your .git/config. This also deletes the directory on disk, AFAIK (although I've never actually used this command myself, so YMMV). (As a side note, you can deinit [and then later re-init] submodules which are still listed in .gitmodules as well if you like, if you just don't want to have a copy of them checked out. This is not considered a change that will push upstream. Of course, if the root build scripts or other dependencies still want them to exist, then you won't actually be able to build in that state.)

-----Original Message----- From: Boost On Behalf Of Gavin Lambert via Boost Sent: 4 July 2019 01:34 To: boost@lists.boost.org Cc: Gavin Lambert Subject: Re: [boost] Link problem with graph and pending still appears unresolved? On 4/07/2019 07:35, Andrey Semashev wrote:

On 7/3/19 6:48 PM, James E. King III wrote:

...

On Wed, Jul 3, 2019 at 11:26 AM Andrey Semashev wrote:

...

Doesn't the disjoint_sets git submodule need to be removed from the superproject?

It was removed from develop in https://github.com/boostorg/boost/pull/283

Strangely enough, it is still present in my .git/config, even after `git pull` and `git submodule update` in the Boost root.

What exists in your config file is as a result of the submodule init commands, and does not get cleared because upstream removed the submodule.

The thing to check whether upstream still wants the submodule to exist or not is the contents of the .gitmodules file.

Once you have confirmed that the module is no longer listed in .gitmodules, run:

git submodule deinit module-name

to remove it from your .git/config. This also deletes the directory on disk, AFAIK (although I've never actually used this command myself, so YMMV).

(As a side note, you can deinit [and then later re-init] submodules which are still listed in .gitmodules as well if you like, if you just don't want to have a copy of them checked out. This is not considered >a change that will push upstream. Of course, if the root build scripts or other dependencies still want them to exist, then you won't actually be able to build in that state.)

I have no mention of disjoint in .gitmodules, (so the remote Boost repos has disjoint_sets removed as noted above.

...

It was removed from develop in https://github.com/boostorg/boost/pull/283

but I DO still have a folder I:\boost\.git\modules\disjoint_sets so are you suggesting git submodule deinit disjoint_sets from the /boost folder? Followed by a 'normal' pull update

git checkout develop git pull --recurse-submodules git submodule update --init

Just checking before I screw up git, again... Paul

On 7/10/19 5:38 PM, Paul A Bristow via Boost wrote:

...

...

git checkout develop git pull --recurse-submodules git submodule update --init

Just checking before I screw up git, again...

OK I've tried this command

I:\boost>git submodule deinit disjoint_sets error: pathspec 'disjoint_sets' did not match any file(s) known to git

but I still have a folder I:\boost\.git\modules\disjoint_sets containing the usual submodule git folders and files.

Should I just delete it? Or leave well alone?

You can delete .git/modules/disjoint_sets and [submodule "disjoint_sets"] from .git/config.

-----Original Message----- From: Boost On Behalf Of Andrey Semashev via Boost Sent: 10 July 2019 16:14 To: boost@lists.boost.org Cc: Andrey Semashev Subject: Re: [boost] Link problem with graph and pending still appears unresolved?

On 7/10/19 6:13 PM, Andrey Semashev wrote:

...

On 7/10/19 5:38 PM, Paul A Bristow via Boost wrote:

...

...

...

git checkout develop git pull --recurse-submodules git submodule update --init

Just checking before I screw up git, again...

OK I've tried this command

I:\boost>git submodule deinit disjoint_sets error: pathspec 'disjoint_sets' did not match any file(s) known to git

but I still have a folder I:\boost\.git\modules\disjoint_sets containing the usual submodule git folders and files.

Should I just delete it? Or leave well alone?

You can delete .git/modules/disjoint_sets and [submodule "disjoint_sets"] from .git/config.

That is, the whole [submodule "disjoint_sets"] section.

That seems to do the trick. I suspect that if I had done boost>git submodule deinit disjoint_sets before deleting anything else, it would have worked as expected. Thanks - all a bit scary 😉 Paul