This is the first release for 2017, following the last release ~4 months ago.
Lots of bug fixes and updates to this release:
-
fixed a progress bar overflow: This bug was reported in duplicate issues:
- #177 reported by @santirodrigo
- #708 reported by @asd-and-Rizzo
- #758 reported by @malteseunderdog
- #765 reported by @malteseunderdog
The problem was that for deletions, we were always deriving the file size from the Source on disk.
The above issues were fixed by PR #784.
-
fixed up spurious pagination that resulted from asynchronous pagination before children were produced.
This issue was reported by @bjohas in #724.
It was fixed by PR #768. -
unshare no matches found bug when no --type was specified.
This issue was reported by @bjohas in issue #747.
The problem was that if a user made a request such as:$ drive unshare -emails somebody@gmail.com -id some_id
without having specified the
--type
, no default accountType was set, hence the
user would get back an non-intuitive/cryptic errorno matches found!
The change in PR #771 fixed it; it adds User
as the defaultType. -
fixed a regression in which emails were no longer being set. This issue was noticed in #762. It was fixed by PR #763.
-
fixed a bug in which local and remote dirTypes differed. There was a long standing
TODO
in code
to fix this issue and it was high time that the chickens had come home to roost. This issue was fixed
by PR #770.
The result of the fix is that if the types differ, we now get an informative error
$ drive new --mime-key docs share_test
$ mkdir -p share_test
$ drive pull share_test; echo $?
Resolving...
/share_test: local is a directory while remote is a file
8
-
fixed up a README inconsistency in the documentation of the various options. This issue was reported by
@jean-christophe-manciot in issue #772. The issue involved
contributions from @sselph and @shaggytwodope. The cause of confusion is that Go's flag package
accepts both-flag
and--flag
indiscriminately. For folks used to *NIX CLI options get confused
because they are accustomed to-s
and--short
instead where:
+-s
is the short option
+--short
is the long option
The issue was fixed by @shaggytwodope with PR #779. -
added the ability to structure .driverc files. This feature was requested by @jean-christophe-manciot in
issue #778. The need for this feature is that using a single global
configuration for every option is very limiting, because one will want--depth=1
for list and--depth=-1
for push, of which the old style of a single configuration style couldn't accomodate.
The solution was implemented by PR #813 and
package https://github.com/odeke-em/namespace.
The new style .driverc file allows for structure e.gid=true [global] depth=10 [pull/list] depth=2 [push] verbose=true [list] long=true
where a section is a command's verb that can be run on the commandline e.g push, pull, list, delete, trash etc, as well as special global
which indicates the fallback/general-purpose section.
The rules of resolution are:
a) If an option is specified in a section that isn't [global]
, it has precedence over the
[global]
or default options e.g
[push]
no-clobber=true
[global]
no-clobber=false
no-clobber=true
wins
b) for options in the same section, the later the option the higher the overriding precedence e.g
[list]
long=false
depth=-1
long=true
In this case long=true
was declared later so long=true
wins
c) multiple sections can be specified in a section, but separated by "/":
[push/pull/list]
depth=-1
no-clobber=true
will make depth=-1
, no-clobber=true
for push, pull and list.
-
fixed an off by one traversal depth bug for md5sum and stat. This bug was reported by
@jean-christophe-manciot in #783. The problem was that at
the entry of the recursive function, we'd asap check ifdepth=0
and exist. However, if we are traversing
a directory/folder, in order to stop recursion on every step we'll decrement the depth by 1. By the time we
hit a non-directory after starting from a directory, we'd not stat/md5sum it. The fix was to first perform
the desired operations and then only check that condition at the end. It was fixed by PR
#832. -
Updated the README documentation to describe how to use
--notify
for file sharing. This issue was reported by @bjohas in issue #791. It was also fixed by @bjohas
with PR #791. -
fixed an illogical return status code on stat failure. This bug was reported by @aueuaio in issue #801. The problem was that the stat error was getting shadowed and also it was being logged to stdout instead of being recomposed and propagated on exit. It was fixed by PR #804.
-
added a new mode to fix clashes with option
--fix-mode
. This feature was designed and implemented by @iblis17 in PR #810. The feature is a new policy that allows you:- trash clashing files.
The default behavior is to rename them.
We can now do
$ drive clashes --fix --fix-mode trash
In that PR, also @jpambrun gave an alternate view and caution on user expectations.
This is pretty cool, it was @iblis17's first Go code and they knocked it out of the park!!
Congrats @iblis17, we are lucky to have you in the drive community and also in the Go community! - trash clashing files.
-
fixed spurious rename error for non-existent-to-begin-with-file. This error was reported in issue #826. The problem and use case was that I was manipulating
files that only exist on the cloud and not on my local disk because I have no local disk space. However,
on performing a rename of a file that only exists remotely, I'd get a spurious error such as
$ drive rename WINTER2017/h1.s binary.s
rename /Users/emmanuelodeke/odeke@ualberta.ca/ENGINEERING/WINTER2017/h1.s /Users/emmanuelodeke/odeke@ualberta.ca/ENGINEERING/WINTER2017/binary.s: no such file or directory
This issue was fixed by PR #827.
-
introduced a debug print mode.
This feature was implemented in PR #830.
Using environment flagDRIVE_DEBUG=true
(carefully chosen to avoid clashing with
other programs that might rely onDEBUG
), one can get debug information dumped to standard output.
The main purpose of this feature is to aid in easy debugging and particularly to alleviate
#829 in which the reporter could have been informed ASAP
that their directory had a .gd directory instead of them having to use their mental microscope
to figure out what went wrong.Typical output will look like
$ DEBUG=true drive list share-testing/ [/Users/emmanuelodeke/go/src/github.com/odeke-em/drive/cmd/drive/main.go:main.discoverContext:1857] contextPath: /Users/emmanuelodeke/emm.odeke@gmail.com/share-testing [/Users/emmanuelodeke/go/src/github.com/odeke-em/drive/cmd/drive/main.go:main.discoverContext:1867] driveRoot: "/Users/emmanuelodeke/emm.odeke@gmail.com" relToRoot: "" [/Users/emmanuelodeke/go/src/github.com/odeke-em/drive/src/rc.go:github.com/odeke-em/drive/src.ResourceMappings:92] RCPath: /Users/emmanuelodeke/emm.odeke@gmail.com/share-testing/.driverc [/Users/emmanuelodeke/go/src/github.com/odeke-em/drive/src/rc.go:github.com/odeke-em/drive/src.ResourceMappings:108] parsedContent from "/Users/emmanuelodeke/emm.odeke@gmail.com/share-testing/.driverc" { "global": { "depth": -1 } } [Commands.List] #0 "/share-testing" -- owner 175.00B 0Bwu8laYc9RTPa28zVk9Td2hTVWc 2016-06-30 15:32:25 +0000 UTC /share-testing/SciqPCKrhi.go -- owner 309.00B 0Bwu8laYc9RTPTXRYblNqQXBSQzQ 2016-02-03 08:12:15 +0000 UTC /share-testing/outf.go -s owner 39.70KB 0Bwu8laYc9RTPOVNSeElpdFBpS2M 2012-02-02 12:00:00 +0000 UTC /share-testing/ComedyPunchlineDrumSound.mp3
Binaries
$ make
CGO_ENABLED=0 GOOS=linux GOARM=5 GOARCH=arm go build -o ./bin/drive_armv5 ./cmd/drive
CGO_ENABLED=0 GOOS=linux GOARM=6 GOARCH=arm go build -o ./bin/drive_armv6 ./cmd/drive
CGO_ENABLED=0 GOOS=linux GOARM=7 GOARCH=arm go build -o ./bin/drive_armv7 ./cmd/drive
CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -o ./bin/drive_armv8 ./cmd/drive
CGO_ENABLED=0 GOOS=darwin go build -o ./bin/drive_darwin ./cmd/drive
CGO_ENABLED=0 GOOS=linux go build -o ./bin/drive_linux ./cmd/drive
rm -f ./bin/md5Sums.txt
find ./bin -type f -name "drive_*" -exec md5 {} >> ./bin/md5Sums.txt \;
cat ./bin/md5Sums.txt
MD5 (./bin/drive_armv5) = 990d899bdff2e54cd401ece624a4f8be
MD5 (./bin/drive_armv6) = 812689b6bb2113948c8394854bbde47f
MD5 (./bin/drive_armv7) = 222be8926c35353f81baba014d926c71
MD5 (./bin/drive_armv8) = ecc56e8df5af018eab40097d261e862d
MD5 (./bin/drive_darwin) = 4b2528c13a5fb955bf28f48beec197ca
MD5 (./bin/drive_linux) = a192d38419377aa27a45d90055c56a14
This release as mentioned is the first of a new year 2017! Hoping for brighter days and more stability, and more user engagement. Our customers, the drive community deserve to have the best commandline client
for Google Drive and that's a constant that we all strive for!
Thank you very much everyone for your patience, contributions, analysis, encouragement, critic, ideas etc.
In particular, I'd like to give a shout to @jean-christophe-manciot for finding many bugs and for all the feedback!
This is a dawn of a new era, let's go for it!! Happy New Year!