pypy3: bump to 5.10.1, fix issues #1475
Conversation
- address 'port lint' warning - properly suppress embedded dependencies Closes: https://trac.macports.org/ticket/55679
|
Notifying maintainers: |
| @@ -35,7 +35,7 @@ patchfiles darwin.py.diff \ | |||
| ffiplatform.py.diff | |||
|
|
|||
| subport pypy3 { | |||
| revision 1 | |||
| version 5.10.1 | |||
There was a problem hiding this comment.
When a port that has subports, especially when the subports have a different version than the main port, it's a good idea to keep a revision line in each subport, and in the main port's if {$subport eq $name} block, even if it's 0.
| set default_prefix "/opt" | ||
| append default_prefix "/local" | ||
|
|
||
| reinplace "s|('/sw/', '${default_prefix}/')|('__PREFIX__',)|" \ |
There was a problem hiding this comment.
Instead of weird hacks, I would recommend handling this as we usually do, with a standard patchfile for extbuild.py that replaces the undesired prefixes with the string __PREFIX__. The reinplace below will then change that back to the desired prefix. In fact, the port used to do it that way, until you changed it in 238d27b. I would change it back.
| @@ -143,6 +148,11 @@ destroot.args --builddir ${destroot}${prefix}/lib \ | |||
| destroot.target package.py | |||
| destroot.post_args | |||
|
|
|||
| subport pypy3 { | |||
| # we set destroot.args just above, so we cannot do this earlier | |||
| destroot.args-append --no-embedded-dependencies | |||
There was a problem hiding this comment.
Is there a reason why the "above" code couldn't also append to destroot.args, rather than overwriting it?
If there is, then this untidiness might be avoided by moving the first subport pypy3 block below where destroot.args is set, so that all of the commands relating to the subport are in one place.
| @@ -143,6 +148,11 @@ destroot.args --builddir ${destroot}${prefix}/lib \ | |||
| destroot.target package.py | |||
| destroot.post_args | |||
|
|
|||
| subport pypy3 { | |||
There was a problem hiding this comment.
The subport procedure is intended to define a subport. This subport has already been defined above; it should not be redefined here. Instead, if this block must really be kept separate from the earlier subport pypy3 block, introduce this block with if {$subport eq "pypy3"} instead.
| @@ -66,12 +65,18 @@ long_description \ | |||
| instead of CPython is speed, as it runs generally faster. | |||
|
|
|||
| post-patch { | |||
| reinplace "s+('/sw/', '/opt/local/')+('__PREFIX__',)+g" \ | |||
| # the PyPy sources search for dependancies in /opt/local and /sw, | |||
| if { ![catch {exec grep -lwre "/sw" -e "/opt/local" ${worksrcpath}}] } { | ||
| return -code error "didn't catch all references to /sw or /opt/local!" | ||
| if { ![catch {exec grep -lwre "/sw" -e ${default_prefix} ${worksrcpath}}] } { | ||
| return -code error "didn't catch all references to /sw or ${default_prefix}!" |
There was a problem hiding this comment.
Is this block really necessary? Have the developers really been making so frequent and sweeping changes to their code that this has actually been an issue? We have tons of other ports that use the patch-with-placeholder-then-reinplace strategy that haven't felt the need to add such sanity checks.
|
Travis Build #1788 Errored. Lint resultsPort pypy success on xcode9.3beta. Log The build timed out. |
|
@danchr Could you fix the things that @ryandesign pointed out above so we can merge this? |
|
@danchr Any progress on fixing the issues that @ryandesign noted? It would be nice to commit this. |
|
@ryandesign, I think @danchr isn't going to show up again for a while. Do you want to add these fixes yourself so we can commit and close this, or should we just close this out until he returns? |
|
I don't have time to fix every problem. Literally anybody else with access could make the changes. |
|
Right, but others don't necessarily have time either, or at least, no one else has shown up with time (and I have fairly little myself). It would be optimal if @danchr made them but he seems to have vanished for the moment. Do you mind if I close this for now? We can reopen it at will. |
|
Sorry, but I just haven’t had the time lately! I should be able to take a look this weekend, but I can’t promise anything… |
|
@danchr If you don't have time, would you prefer to close this until you have time again? |
|
Sure! |
* Bump to latest version, as mentioned. * Address issues identified by Ryan Schmidt in the PR referenced below, such as too much cleverness with replacements. * Properly suppress dependency embedding -- use ours instead. * Reorganise the port to avoid duplication wherever possible. * Add manual page. * Add some comments. * Use pypy2 for building pypy3. * Prevent tkinter from using the system-provided Tk. Closes: https://trac.macports.org/ticket/55679 Refs: #1475
* Bump to latest version, as mentioned. * Address issues identified by Ryan Schmidt in the PR referenced below, such as too much cleverness with replacements. * Properly suppress dependency embedding -- use ours instead. * Reorganise the port to avoid duplication wherever possible. * Add manual page. * Add some comments. * Use pypy2 for building pypy3. * Prevent tkinter from using the system-provided Tk. Closes: https://trac.macports.org/ticket/55679 Refs: #1475
Closes: https://trac.macports.org/ticket/55679
Description
Type(s)
Tested on
macOS 10.13.3
Xcode 9.2
Verification
Have you
port lint?sudo port test?sudo port -vst install?