Opened 17 months ago

Closed 17 months ago

Last modified 15 months ago

#60520 closed defect (fixed)

py36-scipy @1.4.1 build fails on Mac OS Catalina 10.15.4

Reported by: odysseus9672 (Sean Lake) Owned by: michaelld (Michael Dickens)
Priority: Normal Milestone:
Component: ports Version: 2.6.2
Keywords: Cc: chrstphrchvz (Christopher Chavez)
Port: py-scipy

Description (last modified by odysseus9672 (Sean Lake))

I recently upgraded a laptop to Mac OS 10.15.4, and decided to do a clean install of MacPorts. Attempting to run sudo port -s install py36-scipy gives me an error message. Log attached. It looks like it's running into a bunch of errors compiling arpack files, with a bunch of lines like:

:info:build scipy/sparse/linalg/eigen/arpack/ARPACK/SRC/znaitr.f:666:35:
:info:build   383 |             call dvout (logfil, 1, rnorm, ndigit,
:info:build       |                                   2
:info:build ......
:info:build   666 |             call dvout (logfil, 2, rtemp, ndigit,
:info:build       |                                   1
:info:build Error: Rank mismatch between actual argument at (1) and actual argument at (2) (scalar and rank-1))

Similar errors happened when I tried to build arpack @3.7.0 (I thought it was a possible missing dependency).

The XCode version installed is: 11.4.1.

Attachments (1)

main.log (551.2 KB) - added by odysseus9672 (Sean Lake) 17 months ago.
Build log

Download all attachments as: .zip

Change History (10)

Changed 17 months ago by odysseus9672 (Sean Lake)

Attachment: main.log added

Build log

comment:1 Changed 17 months ago by odysseus9672 (Sean Lake)

Description: modified (diff)

comment:2 Changed 17 months ago by ryandesign (Ryan Schmidt)

Owner: set to michaelld
Port: py-scipy added; py36-scipy removed
Status: newassigned

comment:3 Changed 17 months ago by michaelld (Michael Dickens)

Yup ... I see this issue too when building on 10.14. I wouldn't have committed this change if it didn't build using default variants. So ... guessing that something changed in a dependency. Not that that helps much here. It's a Fortran issue ... hmmm ....

comment:4 Changed 17 months ago by michaelld (Michael Dickens)

Resolution: fixed
Status: assignedclosed

In 6e5199719a2b2e92e733ae4f068bf8138888820c/macports-ports (master):

py-scipy: allow legacy Fortran code

Closes: #60520

comment:5 Changed 17 months ago by michaelld (Michael Dickens)

With the move to GCC10, Fortran is more strict about argument types: [20b350bb5eb864e9323f2f79324a2f7f977942d1/macports-ports]. Enabling legacy code resolves the issue for me.

Last edited 16 months ago by ryandesign (Ryan Schmidt) (previous) (diff)

comment:6 Changed 16 months ago by posita (Matt Bogosian (formerly @mbogosian))

This looks to still be an issue with py27-scipy@1.2.3 (which is—as of this writing—a dependency of gqrx@2.12.1_2). Manually doing sudo port install … py27-scipy +gcc9 gqrx && sudo port unsetrequested py27-scipy seemed to work for my case, but maybe that means that gcc10 should be removed as a variant/default dependency if this fix can't be back-ported to this package?

comment:7 Changed 15 months ago by chrstphrchvz (Christopher Chavez)

For reference, the upstream issue: It was addressed by scipy 1.5.0.

comment:8 Changed 15 months ago by chrstphrchvz (Christopher Chavez)

Cc: chrstphrchvz added

comment:9 Changed 15 months ago by chrstphrchvz (Christopher Chavez)

In 184de7671219c609a94487104b7e14ead7dc6453/macports-ports (master):

py-scipy: move gfortran10 fix

Apply workaround for legacy Fortran code to py27-scipy 1.2.3
and py35-scipy 1.4.1, while removing it for latest version
(addressed upstream as of 1.5.0)
See: #60520

Note: See TracTickets for help on using tickets.