Contribute to Open Source. Search issue labels to find the right project for you!

CKEditor Changes Table Style and Border


Are you reporting a feature or a bug?


Check if the issue is already reported

Not Found

Provide detailed reproduction steps (if any)

  1. Go to
  2. Edit <table> tag in “Source” view - edit to resemble the below:

    <table align=“right” border=“0” cellpadding=“5” cellspacing=“0” style=“border: 0px solid #000000;”>

  3. Switch to WYSIWYG view; it has a border.

  4. Switch back to code view: It is now:

    <table align=“right” border=“1” cellpadding=“5” cellspacing=“0”>

Expected result

It shouldn’t put any border, since no border is requested.

Actual result

It thinks a border was requested and sets border=“1”

Other details

  • Browser: Chrome 59 @ Linux, Mac, Windows; Edge
  • OS: Linux, Mac, Windows
  • Installed CKEditor plugins: standard presets


First time bug-poster, long-time user!

Behavior is the same in the nightly build standard editor.

This can be worked-around by breaking the border: tag apart into border-color, border-style, and border-width, which then doesn’t change the border attribute.

Updated 27/06/2017 00:16 2 Comments

password_change not working


Attempted the similar execution as is in the screenshot for #8549 but the arguments logic fails: meterpreter > password_change -u sadmin -oh 31d6cfe0d16ae931b73c59d7e0c089c0 -np ASDqwe123 [-] At least one of -op and -oh must be specified. [-] At least one of -np and -nh must be specified. or

meterpreter > password_change -oh 31d6cfe0d16ae931b73c59d7e0c089c0 -nh 88e4d9fabaecf3dec18dd80905521b29 -u sadmin
[-] At least one of -op and -oh must be specified.
[-] At least one of -np and -nh must be specified.


meterpreter > password_change -op "" -nh 88e4d9fabaecf3dec18dd80905521b29 -u sadmin
[-] At least one of -op and -oh must be specified.
[-] At least one of -np and -nh must be specified.


meterpreter > password_change -oh 31d6cfe0d16ae931b73c59d7e0c089c0 -op "" -np ASDqwe123 -nh 88e4d9fabaecf3dec18dd80905521b29 -u sadmin
[-] At least one of -op and -oh must be specified.
[-] At least one of -np and -nh must be specified.
Updated 26/06/2017 22:06 6 Comments

Make rubocop compliant


There are several steps to this issue:

  1. Add rubocop requirement to Gemfile for test and development environment.
  2. Write a reasonable rubocop.yml file to validate only the code we have written (in app/ and in tests/ etc, but not in lib/). example This can be done as one commit. The second changeset can be:
  3. Fixing the errors we have. This is easy but time consuming.
  4. The last step would be adding a check to CircleCI, which I will do after the above two changesets are added.
Updated 26/06/2017 08:45 1 Comments

doc/ unclear


This is a documentation ticket. I noticed that the docs on “LXD REST API’s security done right” is a bit sparse, spread over multiple files and still requiring to delve deep to understand its possibilities and caveats. Assuming a server with lxd called mightyserv and a devs notebook with lxd called devnb and a setup along the following:

### setup mightyserv
mightyserv # lxc config set core.https_address # listen on a nonstandard port
mightyserv # lxc config set core.trust_password MYSECRETPASS
### review mightyserv
mightyserv # lxc config get core.https_address
mightyserv # lxc config get core.trust_password
# setup devnb
lxc remote add mightyserv
### review devnb
devnb # lxc remote list
devnb # lxc remote list mightyserv:
devnb # lxc list mightyserv:

i found it hard(er) to find info on the following:

  1. when are certs generated in ~/.config/lxc (on devdb)?
  2. how are these certs generated, how can one force their regeneration (i.e. if the used cipher would develop a major vulnerability)
  3. where are client certificates stored (on mightyserv)?
  4. how do you disable client’s (devnb) to mightyserve, when devdb is identified with its specific certificate? lxc trust remove lxc config trust list
  5. how do you list client certificates? lxc config trust list
  6. why would one not want to open the api port to the world if access to it is protected with a strong cipher ( why bother with a ssh tunnel ( if access to the api should be similarly secure as with key-based ssh access to the server? (apart from the obvious reasons such as doubling the security layers and protecting from dos attacks)
  7. how does the api server prevent dos attacks? are there i.e. some fail2ban rules that can be used, or is there a way to white-list access from certain ips?
  8. what do people actually use in production to secure their lxd-api instances?
  9. how do you disable adding new clients (on mightyserv)
  10. best practices, or what does the developer team use in production (everything from nonstadanard api port to guarding servers with war elephants)

Maybe additions to might be helpful? Or am I missing a document covering the above?

Updated 26/06/2017 02:40 3 Comments

lxc file push -r of CWD results in incorrect file names


Required information

  • Distribution: Ubuntu 16.04
  • Kernel version: 4.4.0-79-generic
  • LXC version: 2.14
  • LXD version: 2.14
  • Storage backend in use: dir

Issue description

When you attempt to run lxc file push -rp ./ CONTAINER/dest all of the filenames at /dest within the container are missing characters at the front of all files and dirs. For instance, a ls of the current working dir on the host: bash ~/github/my-project$ ls -a . .. bin .git .gitignore include init lib LICENSE Makefile points.dat src I am instead seeing /dest within the container with partial file/dir names: bash /point # ls -a . .. CENSE b c clude ints.dat it itignore kefile n If it helps at all, the container image I’m using is images:alpine/3.5.

To reproduce: 1. cd projectdir, where projectdir is a dir with files you’d like to copy into the container 1. lxc launch images:alpine/3.5 test 1. lxc exec test mkdir /dest 1. lxc file push -rp ./ test/dest (adding a trailing / after test/dest doesn’t seem to make a difference) 1. lxc exec test ls -a /dest should result in the output described above

Something to note: if you replace the push step with lxc file push -rp ../projectdir test/dest works great! The issue seems to only occur when ./ is explicitly referenced.

There doesn’t seem to be anything useful in the kernel, container, lxd logs nor the client debug output (though, shout out if you think you need these).

Updated 26/06/2017 02:57

Sync FFI, lisp.Func and instr.Instr names


There is some inconsistencies in lisp functions naming inside Go code.

Should aset be ArraySet or Aset? Initially, ArraySet was chosen due to (subjectively) more descriptive name. Lately, Aset appeared in ffi.go, to mimic Emacs Lisp function names.

To achieve consistency again, same naming rules must apply for each package. Keeping Emacs Lisp original names seems like a best option (i.e. Aset, Aget, …).

Updated 24/06/2017 22:22 2 Comments

log,errors: Option to compress rolled log files


Lumberjack recently added the ability to compress rotated log files using gzip. It’d be useful if we could expose that feature in the Caddyfile with a rotation directive similar to rotate_size (which would be before compression), called rotate_gzip or rotate_compress.

The only compression option is gzip. This is enabled by setting the Compress field to true.

The log and errors directives will use this. We do not need to change the core process log’s settings; right now there’s no way to configure that log anyway (other than its filename).

This enhancement is very simple to implement, so I am reserving this for newcomers to the project!

Updated 24/06/2017 22:07

Support million, thousand etc.


Currently “2 million” or “10 thousand” or “1 million / 10 thousand” are not recognized.

Those should be equivalent to “2 * 1 000 000”, “10 * 1000” etc. i.e. unit-less multipliers.

Updated 26/06/2017 21:30 10 Comments

Document credential retry behavior


As seen in the retry behavior when querying ECS or EC2 credential providers retries and can appear to hang due to the excessive exponential backoff. should be updated to describe the retry behavior so it’s not an undocumented pitfall.

Updated 24/06/2017 19:11 1 Comments

`create` should initialize datasets with a default README file


This file should identify a repository as a datalad one, and should give an uninitiated reader some info on datalad and how to make use of the repo, as a dataset. This is related to #1462

Some inspiration for default README content could be taken from:

Updated 26/06/2017 05:31 1 Comments

Notify user when creating duplicate account


<!— Provide a general summary of the issue in the Title above –>

Detailed Description

<!— Provide a detailed description of the change or addition you are proposing –> Check whether the username exists before trying to insert and show a different popup message in this case.


<!— Why is this change important to you? How would you use it? –> <!— How can it benefit other users? –> Inserting duplicate username into the database will throw an exception because the field is unique. The user should also know that an account under his username was already created.

Possible Implementation

<!— Not obligatory, but suggest an idea for implementing addition or change –> Check whether the username exists in the Student model before trying to save the form. If it does then instead of using student.get_new_student_popup() use student.get_existing_student_popup()

Updated 24/06/2017 05:57

Add a reqwest backend to rustup


rustup currently supports a robust collection of HTTP backends: curl, hyper + native-tls and hyper + rustls. Add a fourth based on reqwest, making sure it has proxy and resume support. This will probably become the default backend.

It should be a matter of following the example in src/download, and adding RUSTUP_USE_REQWEST to src/utils/src/


Updated 25/06/2017 07:39 3 Comments

Fix 'show' when active toolchain is not installed


Here’s what happens when you run ‘show’ with a toolchain that is not installed:

brian@DESKTOP-CODN6ST MINGW64 /c/Users/brian/Documents/dev/
$ rustup override set beta
info: using existing install for 'beta-x86_64-pc-windows-msvc'
info: override toolchain for 'C:\Users\brian\Documents\dev\' set to 'beta-x86_64-pc-windows-msvc'

  beta-x86_64-pc-windows-msvc unchanged - rustc 1.19.0-beta.1 (a87984118 2017-06-06)

brian@DESKTOP-CODN6ST MINGW64 /c/Users/brian/Documents/dev/
$ rustup toolchain remove beta
info: uninstalling toolchain 'beta-x86_64-pc-windows-msvc'
info: toolchain 'beta-x86_64-pc-windows-msvc' uninstalled

brian@DESKTOP-CODN6ST MINGW64 /c/Users/brian/Documents/dev/
$ rustup show
Default host: x86_64-pc-windows-msvc

error: override toolchain 'beta-x86_64-pc-windows-msvc' is not installed

Instead of displaying various information about installed toolchains it immediately errors.

It should instead do something like:

brian@DESKTOP-CODN6ST MINGW64 /c/Users/brian/Documents/dev/
$ rustup override remove
info: override toolchain for 'C:\Users\brian\Documents\dev\' removed

brian@DESKTOP-CODN6ST MINGW64 /c/Users/brian/Documents/dev/
$ rustup show
Default host: x86_64-pc-windows-msvc

installed toolchains

nightly-x86_64-pc-windows-msvc (default)

active toolchain

beta-x86_64-pc-windows-msvc (override set by blah blah blah)
(error: override toolchain 'beta-x86_64-pc-windows-msvc' is not installed)

Fix will be in rustup_cli::rustup_mode::show. Instead of returning the error, print it into the show output.

Updated 26/06/2017 15:12 3 Comments

[TAKEN]: Community/Up-For-Grabs: Switch to new configuration keys


The following keys will be renamed: lxc.limit => lxc.prlimit lxc.devttydir => lxc.tty.dir lxc.utsname => lxc.haltsignal => lxc.signal.halt lxc.rebootsignal => lxc.signal.reboot lxc.stopsignal => lxc.signal.stop lxc.logfile => lxc.log lxc.loglevel => lxc.log.level lxc.syslog => lxc.log.syslog lxc.init_cmd => lxc.init lxc.init_uid => lxc.init.uid lxc.init_gid => lxc.init.gid The old jump table methods in confile.c will be moved to confile_legacy.{c,h} and new methods implemented in confile.c. The old and the new configuration keys need to be supported up until LXC 3.0 and dropped afterwards. If someone wants to work on this, just ask for more instructions. If not I will do it.

Updated 27/06/2017 00:24 6 Comments

Convert tests from using explicit constructors to initializers


Here are all the tests that rely on explicit constructors that don’t have an equivalent in the initializers directory (as of June 21, 2017, there are 132 of these tests). These should be updated to use initializers. That may not be possible yet for all of them; if so, it is sufficient to note why they couldn’t be updated yet and come back to them when that issue is resolved.

Note, this list does not include futures.

Known failure categories (i.e., if a test cannot use initializers because of these reasons, just mark it and move on): - tuple fields - array/domain fields initialized via forall expressions (= [i in colors.domain] i;) - [ ] parallel/forall/vass/alist-error - nested records and classes - secondary initializers - some cases of inserting omitted initialization for fields that rely on prior fields - [ ] functions/iterators/bradc/iterWithOnInGenericConstr - inheritance from generic parent classes - inheritance when any class in the tree does not explicitly define an initializer - but only when –force-initializers is not thrown (though there are some limits there as well, so it might be best to punt anyways. Ask me if you want to try anyways) - [ ] distributions/bradc/assoc/userAssoc-domain-priv-check - [ ] distributions/bradc/assoc/userAssoc-array-domain-zipper - [ ] distributions/bradc/assoc/userAssoc-array-noelt - [ ] distributions/bradc/assoc/userAssoc-array-priv-check - [ ] distributions/bradc/assoc/userAssoc-basic-array - [ ] distributions/bradc/assoc/userAssoc-basic-domain - [ ] distributions/bradc/assoc/userAssoc-dom-operations - [ ] studies/labelprop/labelprop-tweets (depends on the userAssoc helper module) - explicitly initializing fields with ‘new’ - [ ] classes/forwarding/dynDispatchVsForward - doesn’t expect coforall + ons in the initializer body - [ ] functions/vass/ref-intent-bug-2big - [ ] The FTree.chpl helper module for the tests in test/studies/madness/aniruddha/madchap/FTree.chpl

Tests to update:

  • [x] arrays/vass/const-checking.fields
    • but we might need to improve the phase 2 const checking
  • [x] classes/bradc/test-class-init-blc
    • will want to update again, either with noinit or if the insertion of omitted fields changes
  • [x] classes/bradc/bitOpOnClass
    • ditto to classes/bradc/test-class-init-blc
  • [x] classes/claridge/baseConstructorCall
  • [ ] classes/constructors/arrayField-userCtor
  • [ ] classes/deitz/constructors/test_construct1
  • [ ] classes/deitz/constructors/test_construct2
  • [ ] classes/deitz/constructors/test_generic_construct1
  • [ ] classes/deitz/constructors/test_generic_construct2
  • [ ] classes/dinan/override_default_constructor
  • [ ] classes/diten/test_destructor
  • [ ] classes/diten/test_destructor2
  • [ ] classes/diten/test_destructor4
  • [ ] classes/ferguson/const-override-bug
  • [ ] classes/figueroa/RecordConstructor1
  • [ ] classes/hilde/callUserDefaultConstructor
  • [ ] classes/hilde/inheritance/constructors-simple
  • [ ] classes/hilde/inheritance/fields
  • [ ] classes/stonea/recordConstructor
  • [ ] classes/vass/old-destructor-name
  • [ ] compflags/lydia/noUseNoinit/noinitSkippedCase (compopts: 1)
  • [ ] compflags/lydia/noUseNoinit/noinitSkippedCase (compopts: 2)
  • [ ] compflags/lydia/noUseNoinit/noinitSkippedCase (compopts: 3)
  • [ ] compflags/lydia/noUseNoinit/recordHasClassField
  • [ ] distributions/diten/testCyclic
  • [ ] domains/vass/assoc-idxtype-error-2
  • [ ] exercises/boundedBuffer/solutions/boundedBuffer-atomic
  • [ ] functions/bradc/namedArgs/forceNamedArgs-createdummy
  • [ ] functions/bradc/namedArgs/forceNamedArgs
  • [ ] functions/hilde/defaultOf/precedence
  • [ ] functions/iterators/claridge/recursiveIterTypeBug
  • [ ] functions/nspark/generic-varargs1
  • [ ] modules/standard/datetime/testDatetimeTZ
  • [ ] modules/standard/datetime/testTimezone
  • [ ] modules/standard/datetime/testTimezoneConversions
  • [ ] multilocale/vass/this-in-taskfns-in-ctors
  • [ ] optimizations/bulkcomm/asenjo/stencilDR/v2/stencil
  • [ ] parallel/sync/bradc/syncArrInit (skip for now)
    • triggers an issue in the type constructor because the field type is the generic version of range while we try to initialize it with an instantiation.
  • [ ] parallel/sync/bradc/syncArrInit2
  • [ ] parallel/taskPar/rachels/Barrier
  • [ ] reductions/bradc/manual/threeclasstypes-construct
  • [ ] reductions/bradc/manual/threeclasstypes-construct2
  • [ ] reductions/diten/deleteReductionClass
  • [ ] release/examples/benchmarks/lcals/LCALSMain (skip rest for now)
    • Half of the types that use constructors can be updated. Of the other two, one fails due to the bug with new statements, the other I suspect will fail due to the extensive use of select statements.
  • [ ] release/examples/benchmarks/miniMD/miniMD (compopts: 1)
  • [ ] release/examples/benchmarks/miniMD/miniMD (compopts: 2)
  • [ ] release/examples/benchmarks/miniMD/explicit/miniMD
  • [ ] release/examples/benchmarks/shootout/chameneosredux
  • [ ] release/examples/benchmarks/shootout/chameneosredux-fast
  • [ ] release/examples/benchmarks/ssca2/SSCA2_main (compopts: 1)
  • [ ] release/examples/benchmarks/ssca2/SSCA2_main (compopts: 2)
  • [ ] release/examples/benchmarks/ssca2/SSCA2_main (compopts: 3)
  • [ ] release/examples/benchmarks/ssca2/SSCA2_main (compopts: 4)
  • [ ] release/examples/benchmarks/ssca2/SSCA2_main (compopts: 5)
  • [ ] release/examples/primers/learnChapelInYMinutes
  • [ ] release/examples/spec/Classes/simpleConstructors
  • [ ] release/examples/spec/Generics/constructorsForGenericFields
  • [ ] release/examples/spec/Records/recordCreation
  • [ ] release/examples/spec/Records/recordDeinitializer
  • [ ] studies/590o/alaska/graph
  • [ ] studies/amr/advection/amr/AMR_AdvectionCTU_driver
  • [ ] studies/amr/advection/grid/Grid_AdvectionCTU_driver
  • [ ] studies/amr/advection/level/Level_AdvectionCTU_driver
  • [ ] studies/amr/diffusion/grid/Grid_DiffusionBE_driver
  • [ ] studies/amr/diffusion/level/Level_DiffusionBE_driver
  • [ ] studies/graph500/v2/main
  • [ ] studies/hpcc/STREAMS/bradc/stream-block1dpar-arr
  • [ ] studies/hpcc/STREAMS/bradc/stream-block1dpar
  • [ ] studies/hpcc/common/bradc/unit/multipleDomains
  • [ ] studies/madness/aniruddha/madchap/test_diff (skip for now)
    • Updating helper module MRA’s big constructor leads to a segfault in copyPropagation. This one is complicated, because it tries to use helper methods to initialize fields, some of which do interesting array initializations. The helper module FTree’s constructor also has the issue with coforall + ons.
  • [ ] studies/madness/aniruddha/madchap/test_gaxpy
  • [ ] studies/madness/aniruddha/madchap/test_likepy
  • [ ] studies/madness/aniruddha/madchap/test_mul
  • [ ] studies/madness/aniruddha/madchap/test_showboxes
  • [ ] studies/madness/aniruddha/madchap/mytests/par-compress/test_compress
  • [ ] studies/madness/aniruddha/madchap/mytests/par-reconstruct/test_reconstruct
  • [ ] studies/madness/aniruddha/madchap/mytests/par-refine/test_refine
  • [ ] studies/matrixMult/stonea/distributed/luLike
  • [ ] studies/matrixMult/stonea/distributed/simpleBlockDist
  • [ ] studies/parboil/BFS/bfs
  • [ ] studies/shootout/chameneos-redux/bradc/chameneosredux-blc
  • [ ] studies/shootout/chameneos-redux/elliot/chameneosredux-ejr
  • [ ] studies/shootout/chameneos-redux/hannah/chameneos-redux-cas
  • [ ] studies/shootout/chameneos-redux/hannah/chameneos-redux-v1
  • [ ] studies/shootout/chameneos-redux/hannah/chameneos-redux
  • [ ] studies/shootout/chameneos-redux/lydia/chameneos-redux-lydia
  • [ ] studies/shootout/chameneos-redux/lydia/chameneos-redux-waitFor
  • [ ] studies/shootout/chameneos-redux/lydia/chameneos-redux-yield
  • [ ] studies/shootout/spectral-norm/lydia/spectral-norm-barrier
  • [ ] studies/shootout/spectral-norm/sidelnik/spectralnorm
  • [ ] studies/shootout/submitted/chameneosredux
  • [ ] studies/ssca2/rachels/SSCA2_test (compopts: 1)
  • [ ] studies/ssca2/rachels/SSCA2_test (compopts: 2)
  • [ ] studies/ssca2/rachels/SSCA2_test (compopts: 3)
  • [ ] studies/ssca2/rachels/SSCA2_test (compopts: 4)
  • [ ] studies/ssca2/rachels/SSCA2_test (compopts: 5)
  • [ ] studies/ssca2/test-rmatalt/nondet
  • [ ] types/records/const-checking/constructors
  • [ ] types/records/const-checking/modifies-const-fields-of-this
  • [ ] types/records/const-checking/tuples-and-records
  • [ ] types/records/ferguson/copy-init/old-ctor-concrete
    • SKIP this is explicitly testing old behavior. We’ll likely want to just remove this test when constructors are deprecated
  • [ ] types/records/hilde/newUserDefaultCtor
  • [ ] types/records/hilde/callConstructorThruTypeFn
  • [ ] types/records/hilde/newUserDefCtorInFn
  • [ ] types/records/hilde/sharedRefCount2
  • [ ] types/records/hilde/callConstructorThruTypeVar
  • [ ] types/records/hilde/callUserDefaultConstructor
  • [ ] types/records/kbrady/record-in-global-tuple
  • [ ] types/records/noakes/return
  • [ ] types/records/sungeun/classInRecord
  • [ ] types/records/sungeun/constructor1
  • [ ] types/records/sungeun/ctorWithOnFn
  • [ ] types/typedefs/lydia/def-new-on-type-alias
  • [ ] types/void/void-variable-def
  • [ ] users/bugzilla/bug794133/bug794133
  • [ ] users/ferguson/refcnt
  • [ ] users/ferguson/recordConstructor
  • [ ] users/npadmana/twopt/do_smu_aos
  • [ ] users/npadmana/twopt/do_smu_soa
  • [ ] users/shetag/tensor-workaround
Updated 26/06/2017 16:34

Greek, Armenian and other numbered list styles not displayed


Are you reporting a feature or a bug?


Provide detailed reproduction steps (if any)

While browsing numbered list style plugin I have found a bunch of unused list styles. The thing is that Array.concat doesn’t add them to the context array, but returns instead.

After changing it in a way it should be so:

if ( ! || CKEDITOR.env.version > 7 ) {
    listStyleOptions = listStyleOptions.concat( [
        [ lang.armenian, 'armenian' ],
        [ lang.decimalLeadingZero, 'decimal-leading-zero' ],
        [ lang.georgian, 'georgian' ],
        [ lang.lowerGreek, 'lower-greek' ]
    ] );

The translations and everything is there in place, it’s just that this feature has not been shown for years.

Custom ordered list types

The thing to ask here is whether we want to include it, or do we want to completely remove it.

I checked our old Trac and it did not include any tickets requesting for this feature, so I say it’s useless and can be safely removed not to bloat the editor.

Updated 23/06/2017 14:23

Tells me to just run ninja but that fails


Using meson-0.41.1-1.fc26.noarch, I’m trying to port totem to meson:

After a few build attempts and thinking my jhbuild gobject-introspection is broken, I run jhbuild uninstall gobject-introspection. Now running ninja says: ninja: error: '/home/hadess/Projects/gnome-install/bin/g-ir-scanner', needed by 'src/Totem-1.0.gir', missing and no known rule to make it

Fair enough, it’s not there any more. Let’s run meson for it to find the dependencies again: ``` Error during basic setup:

Trying to run Meson on a build directory that has already been configured. If you want to build it, just run your build command (e.g. ninja) inside the build directory. Meson will autodetect any changes in your setup and regenerate itself as required.

If you want to change option values, use the mesonconf tool instead. ```

I probably shouldn’t have to nuke my build directory to get out of this loop.

Updated 26/06/2017 12:29 8 Comments

Include course and gender at signup


<!— Provide a general summary of the issue in the Title above –>

Detailed Description

<!— Provide a detailed description of the change or addition you are proposing –> Require students to submit their course (JMC/Computing) and gender at signup.


<!— Why is this change important to you? How would you use it? –> <!— How can it benefit other users? –> Children should be assigned within at least one parent within their course and gender.

Possible Implementation

<!— Not obligatory, but suggest an idea for implementing addition or change –> No need to change views or html for this part, only and New ChoiceFields have to be added to the Student table and the new fields have to be added to the SignUpForm for in

If this approach is chosen then the form is generated from the Student model. Find out more here

Updated 22/06/2017 23:04

Always contract to empty interval


If we find an interval that contracts to the empty interval always choose that interval. (Changes in two places in getBestContractionCandidate: - For the first element - For subsequent elements

Add an if- statement that checks if the new interval is 0 (gain is 1?) and just return that, after reinserting the relevant candidates.)

Updated 22/06/2017 13:59

New SV annotations from clinical genetics

  • [ ] decipher
  • [ ] clingen_cgh_benign
  • [ ] clingen_cgh_pathogenic
  • [ ] clingen_ngi
  • [ ] clingen_ngiAF
##INFO=<ID=decipher,Number=1,Type=Integer,Description="The number of occurances of the event in the database">
##INFO=<ID=clingen_cgh_benign,Number=1,Type=Integer,Description="The number of occurances of the event in the database">
##INFO=<ID=clingen_cgh_pathogenic,Number=1,Type=Integer,Description="The number of occurances of the event in the database">
##INFO=<ID=clingen_ngi,Number=1,Type=Integer,Description="The number of occurances of the event in the database">
##INFO=<ID=clingen_ngiAF,Number=1,Type=Float,Description="The frequency of the event in the database">
Updated 22/06/2017 11:51 1 Comments

Inconsistent number indexes for processes?


I am looking at the default numbers indexes of a local i-score tree. The first Scenario is numbered 1, while the first Event, Timenode, and State are numbered 0.

From a user perspective my opinion is that numbering should in general start with 1. What do you think?


Updated 23/06/2017 18:04 2 Comments

Test file paths not expanded in subtests


Hi vscode-go developers! I’ve taken your function for expanding file paths in go test output into my plugin and found a bug there.

Steps to reproduce:

  1. Create test with subtest and error output, i.e.
func TestIt(t *testing.T) {
    t.Run("SubTest", func(tin *testing.T) {
        tin.Error("Hi, i am not expanded")
  1. Launch test.

  2. Look at path in output. It’s not expanded

--- FAIL: TestIt (0.00s)
    --- FAIL: TestIt/SubTest (0.00s)
        goissue_test.go:7: Hi, i am not expanded

Expected: goissue_test.go:7 -> /Users/user/go/src/

Updated 23/06/2017 18:50 2 Comments

storage: Delete PushTxnRequest.NewPriorities flag


When we changed the way priorities worked in PushTxn, we introduced a new flag to indicate whether the command was supposed to use the old rules or the new. This was done to ensure consistency in the pre-propEvalKV world. Now that propEvalKV has landed, we are no longer bound by the extremely strict backwards-compatibility policy for replica commands and we can delete the flag and change the implementation to use the new mode unconditionally.

Updated 21/06/2017 19:41

Probable undefined behaviour when checking empty files



It appears that LiquidHaskell is performing a division by zero when running liquid on empty file

Steps to reproduce

$ touch empty.hs
$ liquid empty.hs
LiquidHaskell Copyright 2013-17 Regents of the University of California. All Rights Reserved.

**** DONE:  A-Normalization ****************************************************

**** DONE:  Extracted Core using GHC *******************************************

**** DONE:  Transformed Core ***************************************************

**** DONE:  Uniqify & Rename ***************************************************

Working -9223372036854775808% []

**** DONE:  annotate ***********************************************************

**** RESULT: SAFE **************************************************************
Updated 21/06/2017 17:14 1 Comments

Remove lisk-js as submodule


Currently Lisk-JS is listed as submodule for lisk core. Lisk-JS is used only in the lisk core tests, therefore we can get rid of it as git submodule and include it into package.json devDependencies. This is best done when also erasing lisk-ui from the submodules.

Updated 26/06/2017 14:18

Error thrown when pressing a "select all" button in a source view


Are you reporting a feature or a bug?


Provide detailed reproduction steps (if any)

  1. Open a sample.
  2. Switch to the source view.
  3. Press “Select all” button.

Expected result

Action should select whole editor’s content.

Actual result

Exception is thrown: SCRIPT438: Object doesn't support property or method 'createTextRange' ckeditor.js (1065,311)

Other details

  • Browser: Edge
  • OS: Windows 10
  • CKEditor version: 4.7.1
Updated 23/06/2017 13:39

Links to type operators are incorrect


A type operator @ inside a signature links to #:@ but it should link to #t:type (@). For example, click the -* in

Updated 21/06/2017 14:46 1 Comments

dist should warn if you have uncommitted changes while creating the tarball

<nirbheek> hadess, dist will always build from the tarball 
<hadess> what
<nirbheek> It'll extract it to a temporary directory then build it there
<hadess> nirbheek, the tarball that doesn't pass "check"?
<hadess> that doesn't seem like the right thing to do either
<nirbheek> The tarball that is created anew from HEAD each time you run dist
<nirbheek> If you have uncommitted changes it won't put those in, fwict
<hadess> nirbheek, this is horrible
<nirbheek> Shouldn't it always build the dist tarball from the changes in git?
<hadess> i've been trying to fix tests that probably worked 20 minutes
<hadess> 20 minutes ago
<nirbheek> It should warn that you have uncommitted changes while doing dist, yes
<hadess> that would be helpful :)
Updated 21/06/2017 14:08

Display file path for scenarios


It would be nice to see the file path of the scenario that I’m working on.

In some applcations the filepath is displayed in the OS top line, so in i-score it could be: i-score - 1.0.0-b23 - "/Users/i_am_a_boy/i-scoreStuff/MyComposition.jsonscore"

Updated 23/06/2017 18:05 1 Comments

dist target removes build directory even if dist failed

def check_dist():
    unpackdir = tempfile.mkdtemp() 
    builddir = tempfile.mkdtemp()  
    installdir = tempfile.mkdtemp()
    ninja_bin = detect_ninja()     

The rmtrees in the finally: block should not be called when returning early.

Updated 21/06/2017 14:12 3 Comments

Exception thrown when typing a text after tabbing into a table cell


Are you reporting a feature or a bug?


Provide detailed reproduction steps (if any)

Note, the refernced manual test is on a built version - it’s improtant to test it with preset, as it does not occurs without a certain plugins.

  1. Open tableselection/manual/editortypes manual test.
  2. Put a collapsed selection inside “Cell 1.1.1” cell.
  3. Press tab key.
  4. Press shift+tab key to get back into a “Cell 1.1.1” cell.
  5. Type “asd” on your keyboard.

Expected result

Cell text is replaced with “asd”, no exception is thrown.

Actual result

Exceptions are thrown:

SCRIPT5007: Unable to get property 'type' of undefined or null reference
ckeditor.js (141,38)
SCRIPT5007: Unable to get property 'getParents' of undefined or null reference
ckeditor.js (135,80)

Other details

  • Browser: IE11, Edge
  • OS: Windows 10
  • CKEditor version: 4.7.1

Reproducible since CKE 4.7.0 so, since tableselection was shipped.

Since I’m reproducing it on IE11 and Edge I assume it’s a general IE issue.

Updated 26/06/2017 10:53

Fork me on GitHub