Re: Build errors. Again.

[Ryan Gonzalez <rymg19@xxxxxxxxx>]

On June 19, 2015 6:50:22 PM CDT, Ori Bernstein <ori@xxxxxxxxxxxxxx> wrote:
>
>
>> On Jun 19, 2015, at 3:39 PM, Ryan Gonzalez <rymg19@xxxxxxxxx> wrote:
>> 
>> So, I tweaked sysselect.sh, and came up with the following patch:
>
><snip>
>
>Huh, have you pulled recently? I should have replaced that with a
>loop...
>
>http://git.eigenstate.org/ori/mc.git/tree/sysselect.sh
><http://git.eigenstate.org/ori/mc.git/tree/sysselect.sh>
>

I *thought* I had pulled...maybe I didn't.

>
>> Now, there was only more error:
>> 
>> ...
>> ../6/6m -I . install.myr
>> ../6/6m -I . clean.myr
>> ../6/6m -I . test.myr
>> ../6/6m -I . main.myr
>> main.myr:102: Undeclared var std.getcwd
>> make: *** [buildmyr] Error 1
>> 
>> However, I simply deleted my installed libstd and it worked.
>> 
>> So, two issues:
>> 
>> 1. sysselect.sh conditionals messed up.
>> 2. An installed libstd affects the build.
>> 
>> I was going to try to fix the latter, but, looking main.c, it seems
>that the install directory is always appended lastly to the incdirs
>list. Weird.
>
>Yeah; The install directory should be overridden.
>
>I'm not sure why your build is missing it -- the 'main.myr' command
>that gets run on my system is this:
>
>	6m	-I ../libregex -I ../libbio -I ../libstd -I /Users/orib/bin/lib/myr
>main.myr
>
>Which definitely includes the previous directories first.

One more thing: the included bootstrap.sh still has the absolute path to your home directory...

>
>> Anyway, now Myrddin builds! Thanks for the help!
>> 
>> 
>> On Fri, Jun 19, 2015 at 5:05 PM, Ori Bernstein <ori@xxxxxxxxxxxxxx
><mailto:ori@xxxxxxxxxxxxxx>> wrote:
>> The way that the build continues after the error says that it's
>trying to run the bootstrap script,
>> which (these days) lives in bld/bootstrap/bootstrap-$SYS.myr, and is
>kicked off whenever mbld
>> fails. (I should probably change that.)
>> 
>> So, first off:
>> 
>> >  ... lots of other output...
>> 
>> The actual mbld failure is probably buried in there -- there were a
>number of changes, so it's probably
>> not recognizing either a new build attribute or target type (IIRC, I
>changed up 'gen' targets a bit recently).
>> 
>> Now, as far as the fallback failing: The syserrno+foo.myr is selected
>from the base name 'syserrno'
>> using sysselect.sh; what happens if you run:
>> 
>>       cd libstd
>>       ../sysselect.sh syserrno
>> 
>> It looks like it's adding a spurious '+x64' to some files, which is
>causing issues. I'm a bit confused about
>> why, because it definitely checks for whether the files exist.
>> 
>> Assuming that gives the wrong file, you can either try fixing it, or
>if you have trouble fixing the hack
>> bit, it's probably not too hard to just edit the bootstrap script to
>hardcode files that actually exist.
>> 
>> (I should probably drop sysselect.sh and just generate the bootstrap
>scripts with hardcoded names;
>> it's been flaky in the past.)
>> 
>> 
>> > On Jun 19, 2015, at 10:46 AM, Ryan Gonzalez <rymg19@xxxxxxxxx
><mailto:rymg19@xxxxxxxxx>> wrote:
>> >
>> > Unable to open file syserrno+linux-x64.myr
>> >
>> 
>> 
>> 
>> 
>> 
>> -- 
>> Ryan
>> [ERROR]: Your autotools build scripts are 200 lines longer than your
>program. Something’s wrong.
>> http://kirbyfan64.github.io/ <http://kirbyfan64.github.io/> 

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.