Rebol3/Ren-C Branch Change Logs


#41

Date: 2017-05-26T04:46:06Z
Author: Graham Chiu
Commit: 84e1b6aa896f7e34c37a9d650acaa6188151233b
Message: Update prot-http.r

remove the check to see if net-log is working to see if it solves the panic

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-84e1b6a
r3-84e1b6a-debug

0.2.40 OSX x64
r3-84e1b6a
r3-84e1b6a-debug
r3-84e1b6a-debug-cpp

0.3.40 Win64 x86
r3-84e1b6a-debug-cpp.exe
r3-84e1b6a-debug.exe

0.4.4 Linux32 x86
r3-84e1b6a

0.4.40 Linux64 x86
r3-84e1b6a
r3-84e1b6a-debug-cpp


#42

Date: 2017-05-26T08:35:38Z
Author: Barry Walsh
Commit: 2dba411f357aa525ea4de5f82dce9bfbc32ea1fc
Message: Return from host-start helper functions fixed (#524)

  • Tightened what to-dir, get-home-path & get-resources-path can return
  • Have to cover that exists? returns word file or dir

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-2dba411
r3-2dba411-debug

0.2.40 OSX x64
r3-2dba411
r3-2dba411-debug
r3-2dba411-debug-cpp

0.3.40 Win64 x86
r3-2dba411-debug-cpp.exe
r3-2dba411-debug.exe

0.4.4 Linux32 x86
r3-2dba411

0.4.40 Linux64 x86
r3-2dba411
r3-2dba411-debug-cpp


#43

Date: 2017-05-26T20:11:32Z
Author: Shixin Zeng
Commit: 9f38464c741175a58cfb8a4da540173d556147f4
Message: Prevent sendmsg/write raising SIGPIPE the TCP socket is closed:

This is observed by @gchiu etc when testing the http server:

the problem is that the web server writes to the client when the
client has closed the connection this happens when the browser
client does a refresh while not having received yet the full
response

so the server code keeps writing to a broken pipe which doesn't seem
to matter in windows but gives this signal in osx so it's seen on
multiple refreshes with a browser client on the web server we saw
the problem too on android but it went away when we trapped the
writes

There are a few ways to prevent it crashing:

  1. installing a signal handler for SIGPIPE
  2. set SO_NOSIGPIPE option on the socket (only supported on some
    platforms)
  3. set MSG_NO_SIGNAL flag on the sendto (only supported on Linux)

This commit tries to do both 2 and/or 3 when supported.

See:

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-9f38464
r3-9f38464-debug

0.2.40 OSX x64
r3-9f38464
r3-9f38464-debug
r3-9f38464-debug-cpp

0.3.40 Win64 x86
r3-9f38464-debug-cpp.exe
r3-9f38464-debug.exe

0.4.4 Linux32 x86
r3-9f38464

0.4.40 Linux64 x86
r3-9f38464
r3-9f38464-debug-cpp


#44

Date: 2017-05-27T01:11:19Z
Author: Graham Chiu
Commit: 21f0ac7a539fe45b4650432533e7280c45fb75b1
Message: added modules and utils to library

So, this means that modules and functions/utilities are now separated.

so, do <changelog> will look at https://github.com/r3n/renclib/blob/master/userutils.reb

and import <pop3> will go to https://github.com/r3n/renclib/blob/master/usermodules.reb

making it easier to separate out modules from utilities.

We will give people a couple of weeks to upgrade before we start to edit these files to categorize them correctly.

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-21f0ac7
r3-21f0ac7-debug

0.2.40 OSX x64
r3-21f0ac7
r3-21f0ac7-debug
r3-21f0ac7-debug-cpp

0.3.40 Win64 x86
r3-21f0ac7-debug-cpp.exe
r3-21f0ac7-debug.exe

0.4.4 Linux32 x86
r3-21f0ac7

0.4.40 Linux64 x86
r3-21f0ac7
r3-21f0ac7-debug-cpp


#45

Date: 2017-05-27T22:06:55Z
Author: Graham Chiu
Commit: c487c06903d06dea772e28c5710d77966946bc32
Message: allow net-log to display server response and body

So, to show how this works …

    >> net-trace on
Net-trace is now on

>> dl-renc
C: GET / HTTP/1.0
Accept: */*
Accept-Charset: utf-8
Host: metaeducation.s3.amazonaws.com
User-Agent: REBOL


Dumping Webserver headers and body
S: name: %/
size: _
date: _
type: file
response-line: "HTTP/1.1 200 OK"
response-parsed: _
headers: make object! [
    [self: Content-Length Transfer-Encoding Last-Modified x-amz-id-2 x-amz-request-id Date x-amz-bucket-region Content-Type Connection Server]
    [
        Content-Length: _
        Transfer-Encoding: _
        Last-Modified: _
        x-amz-id-2: {c3uMFJ4cwMcj4yPllFaemPrkAH3vbc3SyLdfxJSn+lxyVWGHQ4w0g27ZdPIpO4eezSeHR5L56IM=}
        x-amz-request-id: "0EBD97B1452C57C5"
        Date: "Sun, 28 May 2017 00:08:14 GMT"
        x-amz-bucket-region: "us-east-1"
        Content-Type: "application/xml"
        Connection: "close"
        Server: "AmazonS3"
    ]
]
body: => {<?xml version="1.0" encoding="UTF-8"?>
<ListBucketResult xmlns="...
Available OSes:
1: 0.13.2
2: 0.2.40
3: 0.3.1
4: 0.3.40
5: 0.4.4
6: 0.4.40


0.3.1 Y/n/q
1 2017-05-09/21:23:47.000 r3-2cad72a.exe
..deleted..
38 2017-05-27/22:44:11.000 r3-9fd50b6.exe
What filename by number? (38q)
Downloading ...  r3-9fd50b6.exe
C: GET /travis-builds/0.3.1/r3-9fd50b6.exe HTTP/1.0
Accept: */*
Accept-Charset: utf-8
Host: metaeducation.s3.amazonaws.com
User-Agent: REBOL


Dumping Webserver headers and body
S: name: %/travis-builds/0.3.1/r3-9fd50b6.exe
size: 1596942
date: 27-May-2017/22:44:11
type: file
response-line: "HTTP/1.1 200 OK"
response-parsed: _
headers: make object! [
    [self: Content-Length Transfer-Encoding Last-Modified x-amz-id-2 x-amz-request-id Date ETag Accept-Ranges Content-Type Server Connection]
    [
        Content-Length: 1596942
        Transfer-Encoding: _
        Last-Modified: "Sat, 27 May 2017 22:44:11 GMT"
        x-amz-id-2: {UfTtrLdpMXGjCPNPeqYX6XGb7Rb1fBgrDaKdfjNSvf45lEsjZhVBKdjJZFJs+Q38}
        x-amz-request-id: "5FB1893FDEAB70C4"
        Date: "Sun, 28 May 2017 00:08:26 GMT"
        ETag: {"c88b88a1749f60ca85a125ea215e0003"}
        Accept-Ranges: "bytes"
        Content-Type: "application/x-msdos-program"
        Server: "AmazonS3"
        Connection: "close"
    ]
]
S: 12888 binary bytes in buffer ...
Done.

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-c487c06
r3-c487c06-debug

0.2.40 OSX x64
r3-c487c06
r3-c487c06-debug
r3-c487c06-debug-cpp

0.3.40 Win64 x86
r3-c487c06-debug-cpp.exe
r3-c487c06-debug.exe

0.4.4 Linux32 x86
r3-c487c06

0.4.40 Linux64 x86
r3-c487c06


#46

Date: 2017-05-27T19:47:53Z
Author: Shixin Zeng
Commit: 98ef5706f6e34ee44dea7d0e5583c7ddec19011d
Message: Add a function to retrieve executable path

arg[0] is not reliable to retrieve the executable path, because when the
executable in $PATH, arg[0] could be just the file name, instead of a
full/relative path.

This adds a platform specific function to do this. See:

For platforms this function is not implemented, it will fall back to
argv[0].


#47

Date: 2017-05-27T22:27:45Z
Author: Shixin Zeng
Commit: 9fd50b6e6120e496d75d790f8a764eb3c35b7187
Message: Set system/options/boot in case of no argv

host-main.c has code to ignore invalid argv (due to potential shell
bug?), so theoretically, argv could be empty. Try to set options/boot if
exec-path is set

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-9fd50b6
r3-9fd50b6-debug

0.2.40 OSX x64
r3-9fd50b6
r3-9fd50b6-debug
r3-9fd50b6-debug-cpp

0.3.40 Win64 x86
r3-9fd50b6-debug-cpp.exe
r3-9fd50b6-debug.exe

0.4.4 Linux32 x86
r3-9fd50b6

0.4.40 Linux64 x86
r3-9fd50b6
r3-9fd50b6-debug-cpp


#48

Date: 2017-05-28T03:05:15Z
Author: Shixin Zeng
Commit: 53d73b3fb7376adc3b9824b776799bc35102e485
Message: Fix include for HAVE_PROC_PATHNAME

The code is testing a feature, instead of the platform, so include should do the same.

Also converted some tabs to spaces.

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-53d73b3
r3-53d73b3-debug

0.2.40 OSX x64
r3-53d73b3
r3-53d73b3-debug
r3-53d73b3-debug-cpp

0.3.40 Win64 x86
r3-53d73b3-debug-cpp.exe
r3-53d73b3-debug.exe

0.4.4 Linux32 x86
r3-53d73b3

0.4.40 Linux64 x86
r3-53d73b3


#49

Date: 2017-05-28T06:01:09Z
Author: Graham Chiu
Commit: 4c29aaeaf73f7ef606f7d76ed9586aacfef06492
Message: Fixes size? modified?

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-4c29aae
r3-4c29aae-debug

0.2.40 OSX x64
r3-4c29aae
r3-4c29aae-debug
r3-4c29aae-debug-cpp

0.3.40 Win64 x86
r3-4c29aae-debug-cpp.exe
r3-4c29aae-debug.exe

0.4.4 Linux32 x86
r3-4c29aae

0.4.40 Linux64 x86
r3-4c29aae


#50

Date: 2017-05-29T11:10:04Z
Author: Brian Dickens
Commit: e14b5a007ba0e81723ada45674bcc74805355a4a
Message: Git ignore tmp-* in any subdirectory


#51

Date: 2017-05-29T11:15:16Z
Author: Brian Dickens
Commit: 959c1265df9fd99a781337d67f5f60534e8391a5
Message: Change VAL_TIME() to VAL_NANO(), add checking for REB_DATE/REB_TIME

The nebulously named accessor VAL_TIME() didn’t make it very clear
what the time units were, which is nanoseconds. This changes it to
VAL_NANO() and also makes sure in the C++ build that the accessor is
only used on REB_DATE or REB_TIME values.

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-959c126
r3-959c126-debug

0.2.40 OSX x64
r3-959c126
r3-959c126-debug
r3-959c126-debug-cpp

0.3.40 Win64 x86
r3-959c126-debug-cpp.exe
r3-959c126-debug.exe

0.4.4 Linux32 x86
r3-959c126

0.4.40 Linux64 x86
r3-959c126
r3-959c126-debug-cpp


#52

Date: 2017-05-29T15:34:12Z
Author: Barry Walsh
Commit: b95300848d08da96452730658248ce49c8c3db53
Message: Console skin update bug fix (#525)

system/console would no longer update after a change in commit c712df7
Reverted function? to true? check to make console updates work again.

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-b953008
r3-b953008-debug

0.2.40 OSX x64
r3-b953008
r3-b953008-debug
r3-b953008-debug-cpp

0.3.40 Win64 x86
r3-b953008-debug-cpp.exe
r3-b953008-debug.exe

0.4.4 Linux32 x86
r3-b953008

0.4.40 Linux64 x86
r3-b953008
r3-b953008-debug-cpp


#53

Date: 2017-05-29T19:13:59Z
Author: Shixin Zeng
Commit: 2fb644578746e0ebd128fabfd7b34579541bd0a2
Message: Remove-macro does not need to return anything

It could error out when macro is not found, because “func” expects
something to return

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-2fb6445
r3-2fb6445-debug

0.2.40 OSX x64
r3-2fb6445
r3-2fb6445-debug
r3-2fb6445-debug-cpp

0.3.40 Win64 x86
r3-2fb6445-debug-cpp.exe
r3-2fb6445-debug.exe

0.4.4 Linux32 x86
r3-2fb6445

0.4.40 Linux64 x86
r3-2fb6445
r3-2fb6445-debug-cpp


#54

Date: 2017-05-29T19:03:33Z
Author: Shixin Zeng
Commit: eb534638860c3bc7adcd05bca0087e4d4c05fe1c
Message: Prep needs to be after clean

Or “clean” will delete everything that “prep” generates, and “prep” will
retriggered by “top”


#55

Date: 2017-05-29T19:54:43Z
Author: Shixin Zeng
Commit: aa11f7f8147fbe98195f04788c1f9cfb15324326
Message: Speedup binary-to-c by 5 times

++ is a function, and thus could be much slower than “v + 1”, as
illustrated by make-embeded-header.r:

Before this commit:
$ time ./r3 -qs …/src/tools/make-embedded-header.r
------ Building embedded header file
… Embedded sys-core.h

real 0m32.812s
user 0m32.741s
sys 0m0.070s

After this commit
$ time ./r3 -qs …/src/tools/make-embedded-header.r
------ Building embedded header file
… Embedded sys-core.h

real 0m6.516s
user 0m6.476s
sys 0m0.040s


#56

Date: 2017-05-29T17:10:26Z
Author: Shixin Zeng
Commit: 15a2f3e5ebb5028a0039114cc34865b6151702ff
Message: Make it possible to explicitly set optimization level

if OPTIMIZE= is passed to makefile.boot, it will overwrite the
optimization level set by DEBUG=, unless it’s “auto”, so user can do:

$make -f makefile.boot DEBUG=sanitize OPTIMIZE=3

to have optmized build with address sanitizer.

Setting it to “auto” has the same effect as it being omitted.

This also changes the optimization level of bootstrapping r3-make to 2,
because it’s so slow that travis times out sometimes. Because we don’t
debug these binaries (they are not saved), “-O0” doesn’t mean much, and
the optimized version is really the one we should test with.


#57

Date: 2017-05-29T17:56:04Z
Author: Shixin Zeng
Commit: 6e4446004fa1676fdf1954b10e2f046798e49a54
Message: Build using multiple jobs by default


#58

Date: 2017-05-29T19:20:15Z
Author: Shixin Zeng
Commit: f832ce3e8b1cd84cc2cf6302ed76f1f6315f46fc
Message: Default number of jobs to number of processors


#59

Date: 2017-05-29T21:37:42Z
Author: Shixin Zeng
Commit: b52b5b6d6a369d61931cb29998ff951da5d328f6
Message: Move Android builds after Linux ones

It seems that Travis enqueue builds in the order of they are declared in
the matrix, because native builds have a bootstrapping step, it takes
longer to build, so try to queue them first in hope the the whole job
will finish earlier.


#60

Date: 2017-05-29T21:49:35Z
Author: Shixin Zeng
Commit: 2eb618284481cd9406de875d9def75637f7af17b
Message: Move OSX builds to front

They tend to wait longer, probably due to limited OSX resource
available.

Move them earlier in hope that they can start waiting earlier

The binaries below are only available for a couple of weeks or so after commit date.

0.13.2 Android5-arm
r3-2eb6182
r3-2eb6182-debug

0.2.40 OSX x64
r3-2eb6182
r3-2eb6182-debug
r3-2eb6182-debug-cpp

0.3.40 Win64 x86
r3-2eb6182-debug-cpp.exe
r3-2eb6182-debug.exe

0.4.4 Linux32 x86
r3-2eb6182

0.4.40 Linux64 x86
r3-2eb6182
r3-2eb6182-debug-cpp