always allow trying mosek_cli #966
Conversation
|
@whoburg, ready for review. |
|
Failing only because of MOSEK 8. @whoburg this is ready for review. |
|
Still not the right answer. You now find the 32 bit mosek_cli on WIndows 7, but gpkit will fail if it tries to use it. You don't see a failure at the moment because I'm forcing only the cvxopt solver. |
|
re: MOSEK 8, not yet begun. |
|
Oops, I had it backwards. Windows 7 has a 32 bit version of python using pythonxy, but mosek is 64bit. That combination does not work. I never upgraded windows 7 to us anaconda after we realized that you could install cvxopt with anaconda. |
|
oh hmm, so it's py32 sys.exec not working with mosek64. updated the issue accordingly... |
|
Yeah I don't know how the mosek_cli works, but from what I recall the mixture of bits don't play nice with each other. Let me turn of the forcing of cvxopt and see what happens. |
|
test this please |
|
Not the error that I recalled, but this is what you get now: |
|
Hmm! It looks like it successfully ran the |
|
Looks like it's using |
|
Apparently. I don't remember what the problem was, but ill update jenkins
to run mosek_cli on Windows 7.
…On Thu, 01 Dec 2016, 21:42 bqpd ***@***.***> wrote:
Looks like it's using mosek_cli just fine!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#966 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AFmotN_dzKItoRFlNlD6ZlJWZYjTg6Wtks5rD4WggaJpZM4LAzHy>
.
|
|
Test this please |
|
oops, @galbramc, I think it's still finding MOSEK 8, but for mosek_cli only? |
|
@whoburg the code is ready for review, though tests are pending. |
|
retest this please |
gpkit/_mosek/cli_expopt.py
Outdated
| settings["mosek_bin_dir"]) | ||
| if "mosek_bin_dir" in settings: | ||
| os.environ['PATH'] = (os.environ['PATH'] + ':%s' % | ||
| settings["mosek_bin_dir"]) |
There was a problem hiding this comment.
":".join([os.environ['PATH'], settings["mosek_bin_dir"]])
seems cleaner
There was a problem hiding this comment.
Still doesn't fit on one line, sadly.
|
|
||
| def look(self): | ||
| "Attempts to import mskexpopt." | ||
| "Attempts to import cvxopt." |
| try: | ||
| import gpkit | ||
| log("# Moving to the directory from which GPkit was imported.") | ||
| os.chdir(gpkit.__path__[0]) |
There was a problem hiding this comment.
hmm, does this have the potential to cause problems if there are multiple installs of GPkit and the user wants to run the build script to overwrite the currently-installed one?
There was a problem hiding this comment.
sorry, shouldn't have said multiple installs -- rather, envisioning a case where a user has a system-wide install (in Python site-packages) but wants to run the build script from e.g. a clone of the git repo.
There was a problem hiding this comment.
I believe that the install step will remove the already installed one before building, and this passes tests and helps with #959 so I figured it was worth a try.
There was a problem hiding this comment.
ok yeah, sounds like the removal step addresses the concern.
| _repr_latex_ = _repr_latex_ | ||
|
|
||
| def str_without(self, excluded=()): | ||
| def str_without(self, excluded=None): |
There was a problem hiding this comment.
fine, but seems unrelated?
|
Hold on, I'm updating the testing script. |
|
test this please |
|
code is approved but I don't want us to merge until two conditions are both true:
|
|
retest this please |
|
(@galbramc, on unit tests but not dependencies still finding mosek8 on win10) |
|
I just reinstalled mosek 7. |
|
test this please |
|
@bqpd, looks good to merge. |
|
It is done! |
Closes #963
Addresses debugging flow of #959