blockdev- 
 17:01: 
 pVDMA6LsnKslYMZRObRkbTAnXR 
 DEIyisSDyjrHNaAXxV9TroIWKF 
 sO028+drNExJAkMwsAZM5fwloO3 
 vbciEPgzr/207KfxFOoyDVlg9b2 
 M36ZZbT47faKUDNVb+3XQz7AOrW 
 rTU40m3Vkefp38hZWmCepETVziP 
 ICiRkDM25EKMID+/5wu9MhAYe1e 
 00721157ae682- 
 ( 
 From: dankamongmen@gmail.com 
  
 ok, i uploaded growlight 1.2.40 to experimental and have 
 debugging logs in the build now. we're bombing in the same place 
 there, though with a different error code: 
  
 interrogate_terminfo:1395:terminfo error -1 for [] (see terminfo(3ncurses)) 
 Couldn't set up notcurses 
  
 so this suggests we have no TERM defined in buildd. and indeed, 
 if we look earlier: 
  
 test 1 
     Start 1: growlight-tester 
  
 1: Test command: /build/reproducible-path/growlight-1.2.40/obj-x86_64-linux- 
 gnu/growlight-tester 
 1: Working Directory: /build/reproducible-path/growlight-1.2.40/obj-x86_64- 
 linux-gnu 
 1: Test timeout computed to be: 1500 
 test 2 
     Start 2: blockdev-verbose 
 1: Running with LANG=C.UTF-8 
 1: TERM wasn't defined, exiting with success 
 1/2 Test #1: growlight-tester .................   Passed    0.00 sec 
  
 it's expected that notcurses fails to initialize in this case. 
  
 is this something that changed since 1.2.38 in the buildd 
 environment? where could we find that out? 
  
 we could override dh_auto_test to supply TERM if it's not 
 defined, or better yet only in the buildd environment, or better 
 yet not do this unholy thing. 
  
 we could amend the growlight cmake file to require TERM when 
 running tests. but we'd want to return error in that case, which 
 doesn't improve our situation. 
  
 we run these same tests in autopkgtests, explicitly defining 
 TERM there. so we could just bypass this in dh_auto_test, and 
 rely on autopkgtest...i don't love it. surely other people have 
 needed TERM during builds? 
  
 --- SoupGate-Win32 v1.05 
  * Origin: you cannot sedate... all the things you hate (1:229/2)