# Extension build failures

**URL:** <https://discourse.slicer.org/t/extension-build-failures/1274>\
**Category:** Support\
**Tags:** build, extensions-manager, bug, package\
**Created:** [October 24, 2017, 1:55pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274 "2017-10-24T13:55:08Z")\
**Posts on this page:** 13\
**Page:** 1

<div class="post-metadata">

**Author:** ![fedorov](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/fedorov/32/14_2.png) [@fedorov](https://discourse.slicer.org/u/fedorov)\
**Post date:** [October 24, 2017, 1:55pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/1 "2017-10-24T13:55:08Z")

</div>

There are new failures in some of the extensions starting from Oct 19 (sorry, I noticed just now). We did not change the extensions code in that period of time, so it must be due to some Slicer changes.

Specifically, DCMQI and PkModeling started to fail (both are superbuild-style extensions).

Last successful dashboard: [CDash](http://slicer.cdash.org/index.php?project=Slicer4&date=2017-10-17&filtercombine=or&filtercount=2&showfilters=1&filtercombine=or&field1=buildname&compare1=63&value1=DCMQI&field2=buildname&compare2=63&value2=PkModeling)

First time they started to fail: [CDash](http://slicer.cdash.org/index.php?project=Slicer4&date=2017-10-19&filtercount=2&showfilters=1&filtercombine=or&field1=buildname&compare1=63&value1=DCMQI&field2=buildname&compare2=63&value2=PkModeling)

Specific errors:

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/5/4/54221807384cf4860ffed97e23830125a037b2ae.png)

 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/a/a/aa4bf4aa7a5cc59c6b175130f9306307f8a5fc6c.png)

At the same time, the packages appear to be generated.

---

<div class="post-metadata">

**Author:** ![ihnorton](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/ihnorton/32/9_2.png) [@ihnorton](https://discourse.slicer.org/u/ihnorton)\
**Post date:** [October 24, 2017, 3:35pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/2 "2017-10-24T15:35:45Z")

</div>

Also seeing this, but only for Mac build (UKFTractography extension).

---

<div class="post-metadata">

**Author:** ![fedorov](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/fedorov/32/14_2.png) [@fedorov](https://discourse.slicer.org/u/fedorov)\
**Post date:** [October 24, 2017, 3:51pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/3 "2017-10-24T15:51:04Z")

</div>

For the extensions I mentioned, it is also only Mac builds.

---

<div class="post-metadata">

**Author:** ![fedorov](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/fedorov/32/14_2.png) [@fedorov](https://discourse.slicer.org/u/fedorov)\
**Post date:** [October 25, 2017, 3:33pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/4 "2017-10-25T15:33:43Z")

</div>

@jcfr I did not test, but on the surface at least (the error happens in the fixup\_extension step, the specific failure is about not being able to locate the extension dir, the error happens in the superbuild extensions) it looks like this commit is to blame:

[https://github.com/Slicer/Slicer/commit/d6c3a2cf4999c4a3eb5dedd974e1f1aa6fd7aac6](https://github.com/Slicer/Slicer/commit/d6c3a2cf4999c4a3eb5dedd974e1f1aa6fd7aac6)

What do you think?

---

<div class="post-metadata">

**Author:** ![markasselin](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/markasselin/32/5916_2.png) [@markasselin](https://discourse.slicer.org/u/markasselin)\
**Post date:** [October 27, 2017, 1:55am UTC](https://discourse.slicer.org/t/extension-build-failures/1274/5 "2017-10-27T01:55:35Z")

</div>

In the latest Windows nightly (4.9.0-2017-10-25) I don’t see _any_ extensions at all. Below is a screenshot of what I see.

 ![BrokenExtensionManager](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/f/3/f3d3746e55afc64788b4f5f821bb5a00b286959d.jpeg)

---

<div class="post-metadata">

**Author:** ![jcfr](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/jcfr/32/17825_2.png) [@jcfr](https://discourse.slicer.org/u/jcfr)\
**Post date:** [October 27, 2017, 6:01pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/6 "2017-10-27T18:01:54Z")

</div>

> [@fedorov](#):
>
> I did not test, but on the surface at least (the error happens in the fixup\_extension step, the specific failure is about not being able to locate the extension dir, the error happens in the superbuild extensions) it looks like this commit is to blame:

Thanks for the report. If it hasn’t been addressed yet, I will follow up shortly.

---

<div class="post-metadata">

**Author:** ![fedorov](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/fedorov/32/14_2.png) [@fedorov](https://discourse.slicer.org/u/fedorov)\
**Post date:** [November 4, 2017, 3:30pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/7 "2017-11-04T15:30:54Z")

</div>

Unfortunately, the build errors are still there.

---

<div class="post-metadata">

**Author:** ![fedorov](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/fedorov/32/14_2.png) [@fedorov](https://discourse.slicer.org/u/fedorov)\
**Post date:** [November 4, 2017, 3:31pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/8 "2017-11-04T15:31:32Z")

</div>

@jcfr what would be the consequences of reverting the commit that introduced the regression?

---

<div class="post-metadata">

**Author:** ![jcfr](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/jcfr/32/17825_2.png) [@jcfr](https://discourse.slicer.org/u/jcfr)\
**Post date:** [November 4, 2017, 7:51pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/9 "2017-11-04T19:51:59Z")

</div>

Tackling this now, we should have fix in shortly.

---

<div class="post-metadata">

**Author:** ![jcfr](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/jcfr/32/17825_2.png) [@jcfr](https://discourse.slicer.org/u/jcfr)\
**Post date:** [November 4, 2017, 8:05pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/10 "2017-11-04T20:05:11Z")

</div>

Problem identified. PR for PkModeling and DCMQI on the way

---

<div class="post-metadata">

**Author:** ![jcfr](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/jcfr/32/17825_2.png) [@jcfr](https://discourse.slicer.org/u/jcfr)\
**Post date:** [November 4, 2017, 8:51pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/11 "2017-11-04T20:51:36Z")

</div>

PRs fixing DCMQI and PkModeling have been submitted:

- [https://github.com/QIICR/dcmqi/pull/310](https://github.com/QIICR/dcmqi/pull/310)
- [https://github.com/millerjv/PkModeling/pull/65](https://github.com/millerjv/PkModeling/pull/65)

Also, Slicer extension build system has been improved to help extension developer diagnose the problem at configuration time. An configure like the following will now be reported:

```auto
NameOfExtension: Variable CPACK_INSTALL_CMAKE_PROJECTS is expected to be set.

```

[https://github.com/Slicer/Slicer/commit/b160ec13f276a86306513954ef8b08a5332afc2e](https://github.com/Slicer/Slicer/commit/b160ec13f276a86306513954ef8b08a5332afc2e)

---

<div class="post-metadata">

**Author:** ![ihnorton](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/ihnorton/32/9_2.png) [@ihnorton](https://discourse.slicer.org/u/ihnorton)\
**Post date:** [November 6, 2017, 3:01pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/12 "2017-11-06T15:01:00Z")

</div>

Would the value of `CPACK_INSTALL_CMAKE_PROJECTS` ever be different than in those PRs?

---

<div class="post-metadata">

**Author:** ![jcfr](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/jcfr/32/17825_2.png) [@jcfr](https://discourse.slicer.org/u/jcfr)\
**Post date:** [November 6, 2017, 3:53pm UTC](https://discourse.slicer.org/t/extension-build-failures/1274/13 "2017-11-06T15:53:32Z")

</div>

> [@ihnorton](#):
>
> Would the value of CPACK\_INSTALL\_CMAKE\_PROJECTS ever be different than in those PRs?

If the project only includes Slicer modules, it should not be different.

In the case of this PR, we explicitly install the `RuntimeLibraries` component. Other more complex project could have other components …