Bug 222553 - math/R fails to compile when LTO is set
Summary: math/R fails to compile when LTO is set
Status: Closed Overcome By Events
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: Any Any
: --- Affects Some People
Assignee: Joseph Mingrone
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-09-24 12:59 UTC by Grzegorz Junka
Modified: 2017-09-29 18:11 UTC (History)
1 user (show)

See Also:
bugzilla: maintainer-feedback? (jrm)

Attachments
Compilation logs (167.75 KB, text/x-log)
2017-09-24 12:59 UTC, Grzegorz Junka 		no flags Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Grzegorz Junka 2017-09-24 12:59:27 UTC 
Created attachment 186661 [details]
Compilation logs

When LTO is set:

OPTIONS_FILE_SET+=LTO

port fails to compile - see attached log.
Comment 1 Joseph Mingrone freebsd_committer freebsd_triage 2017-09-24 13:39:07 UTC 
Matching the settings in your build, but in a clean environment (poudriere), the build finishes successfully.

http://pkg.awarnach.mathstat.dal.ca/data/11amd64-default/2017-09-24_10h20m09s/logs/R-3.4.1_7.log

Could you try again with the suggestion at the end of your build?  That is, try `make MAKE_JOBS_UNSAFE=yes build`.
Comment 2 Grzegorz Junka 2017-09-24 14:24:59 UTC 
I haven't updated ports for about two weeks. Maybe something is already fixed. I will double check. I will schedule the compilation today and should have the result by tomorrow.
Comment 3 Joseph Mingrone freebsd_committer freebsd_triage 2017-09-25 15:28:09 UTC 
I've been able to reproduce this now.  I am considering just removing the LTO option.  It's not turned on by default upstream and it's been a lot of trouble for questionable benefit.
Comment 4 david 2017-09-25 16:55:55 UTC 
Thank you for this hint: I have been having trouble building math/R under portmaster for the bast few months; when I saw this report, I checked: sure enough, I had the LTO option set.  So I un-set it; this morning, the update of math/R from R-3.4.1_7 to R-3.4.1_8 using portmaster "Just Worked."

(In my case, the apparent symptom seemed to involve the attempted use of an "ar" executable that was from a different toolchain than expected: a command-line option had been specified that the "ar" that was actually invoked failed to understand.)
Comment 5 Joseph Mingrone freebsd_committer freebsd_triage 2017-09-28 12:03:04 UTC 
(In reply to david from comment #4)
David, see https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=220481
Comment 6 Joseph Mingrone freebsd_committer freebsd_triage 2017-09-29 18:11:16 UTC 
Thanks for reporting.  The LTO option has been removed with the upgrade to 3.4.2.