Repeated make code

[ahmadia]

In the messenger directory, the file make.py appears to be copied (with very minor modifications) separately into the Linux and OS X build directories. It's not copied into the Windows directory, making for an inconsistent build experience.

The intention and usage of the make.py bootstrap script should be clarified.

[arokem]

Sorry for the delayed response. Users are not supposed to run these make
scripts. The intention of these scripts is that we ourselves run them on
each of the platforms and provide an already compiled binary. If we change
the name of these scripts would that change your expectations about them
and be less confusing?

On Mon, Apr 14, 2014 at 8:10 PM, Aron Ahmadia notifications@github.comwrote:

In the messenger directory, the file make.py appears to be copied (with
very minor modifications) separately into the Linux and OS X build
directories. It's not copied into the Windows directory, making for an
inconsistent build experience.

The intention and usage of the make.py bootstrap script should be
clarified.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/66
.

[ahmadia]

I'm a "user" in the sense that I need to build these binaries myself. For maintainability purposes, it probably makes sense to only have one of these. I'll submit a PR if I've got some reasonable changes.

[arokem]

Yes - I agree that a reorganization of that code could easily remove some
redundancy.

Why do you have to build these yourself? Can't you just use the binaries we
compiled?

On Sat, Apr 26, 2014 at 6:19 PM, Aron Ahmadia notifications@github.comwrote:

I'm a "user" in the sense that I need to build these binaries myself. For
maintainability purposes, it probably makes sense to only have one of
these. I'll submit a PR if I've got some reasonable changes.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/66#issuecomment-41485668
.

[ahmadia]

@cekees has been recompiling them in our stack. I assume the mex bindings are specific to the libraries being used.

[cekees]

Yes, I wanted to make pymatbridge a fully functional package within
hashdist so I was building the messenger extension so that it points to
libzmq, etc. from the active hashdist profile.

On Sun, Apr 27, 2014 at 12:13 AM, Aron Ahmadia notifications@github.comwrote:

@cekees https://github.com/cekees has been recompiling them in our
stack. I assume the mex bindings are specific to the libraries being used.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/66#issuecomment-41488678
.

[arokem]

What were you doing to build? Were you using our make.py scripts? Was it
fairly straight-forward?

On Mon, Apr 28, 2014 at 7:51 AM, Chris Kees notifications@github.comwrote:

Yes, I wanted to make pymatbridge a fully functional package within
hashdist so I was building the messenger extension so that it points to
libzmq, etc. from the active hashdist profile.

On Sun, Apr 27, 2014 at 12:13 AM, Aron Ahmadia notifications@github.comwrote:

@cekees https://github.com/cekees has been recompiling them in our
stack. I assume the mex bindings are specific to the libraries being
used.

—
Reply to this email directly or view it on GitHub<
https://github.com/arokem/python-matlab-bridge/issues/66#issuecomment-41488678>

.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/66#issuecomment-41567506
.

[cekees]

Yes, the only thing that hasn't been straightforward is the cygwin build,
because I was trying to use gnu compilers and matlab doesn't have defaults
for building mex files with gnu on cygwin, though it can be done.

On Mon, Apr 28, 2014 at 9:53 AM, Ariel Rokem notifications@github.comwrote:

What were you doing to build? Were you using our make.py scripts? Was it
fairly straight-forward?

On Mon, Apr 28, 2014 at 7:51 AM, Chris Kees notifications@github.comwrote:

Yes, I wanted to make pymatbridge a fully functional package within
hashdist so I was building the messenger extension so that it points to
libzmq, etc. from the active hashdist profile.

On Sun, Apr 27, 2014 at 12:13 AM, Aron Ahmadia notifications@github.comwrote:

@cekees https://github.com/cekees has been recompiling them in our
stack. I assume the mex bindings are specific to the libraries being
used.

—
Reply to this email directly or view it on GitHub<

https://github.com/arokem/python-matlab-bridge/issues/66#issuecomment-41488678>

.

—
Reply to this email directly or view it on GitHub<
https://github.com/arokem/python-matlab-bridge/issues/66#issuecomment-41567506>

.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/66#issuecomment-41567736
.

[arokem]

This #73 should help. Could you take a look?

[blink1073]

I'm calling this one fixed by #144. Feel free to submit any cygwin specific changes @cekees.

[cekees]

Thanks. I haven't gotten around to updating pymatbridge in a while. I
appreciate all the effort you guys are pointing into it and will try to get
our stack updated to a recent master at some point in the next month or so.

On Wed, Feb 18, 2015 at 9:44 PM, Steven Silvester notifications@github.com
wrote:

I'm calling this one fixed by #144
#144. Feel free to
submit any cygwin specific changes @cekees https://github.com/cekees.

—
Reply to this email directly or view it on GitHub
#66 (comment)
.