IRC channel logs

2022-05-22.log

back to list of logs

<taw10>Could it be related to using readline?
<taw10>If I move my .guile (containing activate-readline and activate-colorized) away, COLUMNS seems to always be #f
<taw10>Seems like only COLUMNS get special handling. Other variables (PATH, HOSTNAME and even random stuff like JOURNAL_STREAM) gets propagated
<taw10>I can't find any mention of COLUMNS in the Guile source code. Could it be that readline does something strange?
<taw10>Ok, I think I know what's happening. Apart from the ungraceful handling in terminal-width, that is
<taw10>COLUMNS isn't a real environment variable, but rather a kind of pseudo-variable. There's no environment variable to "export" into the Guile process in the first place, which is why COLUMNS is unset. It only starts getting set once activate-readline is called and readline starts taking care of it
<taw10> https://stackoverflow.com/questions/1780483/lines-and-columns-environmental-variables-lost-in-a-script
<taw10>The activate-readline happens quite late in the Guile startup. When I run my test case directly ("guile exception-test.scm"), the program gets run without hitting the REPL at all. When using (load "exception-test.scm"), we're alreay in a readliney situation
<taw10>Ok, bug report submitted. Thanks for the help, stis! This was a nice little puzzle to solve
<taw10>stis: For info, here's my bug report: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=55568
<stis>tahnks taw10, looks good!
***chris is now known as Guest5506
***lampilelo_ is now known as lampilelo
***Guest5506 is now known as chrislck
<chrislck>sneek: botsnack
<sneek>:)