showvikdbz/code-reviews · Datasets at Hugging Face

text stringlengths 1 5.2k	labels listlengths 1 1
this is odd why do you need the why not just env exceptionoccurred or just ifs null is there a case where env exceptionoccurred and ifs null if that is the case you will need to clear the exception before you proceed	[ 0 ]
the formatting looks off please run presubmit sh before submitting	[ 0 ]
nit put this below after private	[ 0 ]
if you are trying to minimize wasted work why waste time on basename getprogname	[ 0 ]
why are these lists not assoc but instead this weird struct-esque structure?	[ 1 ]
Damn I tested the example but didn't copy paste it so didn't catch the space.	[ 1 ]
it s effectively dead code dating from before my redesign to create dashboards from json	[ 0 ]
This looks like a crappy test.	[ 1 ]
This was probably a bad interaction of Olympic network tx spamming and the ridiculous I/O load that we threw at leveldb due to a bug in state DB dirty tracking.	[ 1 ]
fall through the end with self neutron service plugins none which was the original behaviour	[ 0 ]
i cannot quite tell but does not this fit without the escape char port dict self core plugin make port dict port process extensions false	[ 0 ]
> realistic use cases for these things require making copies of that buffer, and then you're screwed again Why are you screwed? As long as you `Py_INCREF` after making that copy and before doing anything with it, you're fine, right? > I did a quick skim of numpy and scipy and every use looks broken to me. When combined with this patch, numpy looks alright to me, in the one place we use it. That doesn't mean this is a good api, but it does look like a usable one.	[ 1 ]
Damn. This crash never showed up in debug mode for some reason but only in release build. Fix [1957eba]	[ 1 ]
hmm returning here unconditionally seems wrong we have not run validatejsonstring for the correctly typed entries after fixing this it would be good to have a unit test for this case	[ 0 ]
My trend is always to avoid Net::HTTP (it's slow, clumsy, and buggy) I use ftw (I wrote it) instead of Net::HTTP because that library is quite horrible Open to accepting patches to add proxy support to ftw ;)	[ 1 ]
should this also pass through the authorization code ttl	[ 0 ]
Hopefully it's better now, but I suck at complex sentences.	[ 1 ]
This sounds very ridiculous, I'm not sure if we really want it?	[ 1 ]
did you try /atom/Click or /atom/proc/Click? using . here would i think fuck you up	[ 1 ]
could use a list of turfs instead of world, pretty much anything "in world" is shit for lag.	[ 1 ]
how about doing this when dealing with the stringfactory you would then only push new instance of strings	[ 0 ]
either const iterator findintervalstart uint t from or std pair bool int t find uint t from the from and to are also bad names but let us not change them in this cl	[ 0 ]
only a stub info are in the wadl file	[ 0 ]
please throw a nullpointerexception	[ 0 ]
realized essex should be removed	[ 0 ]
U fuckin faggot I swear to God I'll find u	[ 1 ]
you could replace your whole logic with a simple for loop size t i for i i count use buf i	[ 0 ]
yes none or insignificant if you prefer importlib or not you need to call shell commands to plug your interface like ovs vsctl in a shell	[ 0 ]
Because breaking functions is a fucking good idea. Can't you keep it on del process and not use qdel?	[ 1 ]
nit false is scan request	[ 0 ]
what do you plan for deprecating the old value in order to avoid confusion for people wanting to update retries count	[ 0 ]
it works in trunk and diablo stable is there a reason you are testing against the rcb build i am not aware of that being a publicly available build or that we need to be testing against it	[ 0 ]
what are you take into account in order to simulation a migration live block dirty page bandwidth	[ 0 ]
get rid of bits in byte you can do div round up add out pixelbits instead	[ 0 ]
Colour is great - thanks I am not a css expert but I am pretty sure that the css can be written in shorthand instead of all those classes being specified directly. @C-Lodder @ciar4n am I correct or talking the usual rubbish	[ 1 ]
the sprite is entirely blob colors, so it'd be a bit weird having dough in the recipe but no dough in the sprite	[ 1 ]
let s eradicate the black magic stdout stderr utils execute chroot chroot dpkg query w f package version pkgver list stdout split	[ 0 ]
In GitLab by @jeremyletang on Nov 21, 2019, 14:29 yes it's nonsense indeed.	[ 1 ]
scratch that i was being an idiot	[ 0 ]
you should absolutely remove the period here plog adds a colon and an error string	[ 0 ]
manager should be removed here	[ 0 ]
makes sense but if that is the kind of restriction someone wants to enforce could not they just use the policies setting instead of the roles i do not mean to be a jerk i am just trying to understand what extending the semantics of the roles style property protections gets you that is not already available from the policies style property protections	[ 0 ]
stupid nit added vs available or accessible from the client perspective maybe	[ 0 ]
You can just call animate() with a 3 like that???!!? wtf byond pls Even if you can, I would probably make it time = 3 just for clarity	[ 1 ]
remove the ifdef here	[ 0 ]
ah i would forgotten about the disabled by file part of the middlware i guess the difference between a graceful shutdown and using the middleware is that one is event driven where the other one requires you to pick a long time and assume that after that there are no requests being handled which would work but the code to implement the graceful shutdown could be pretty easy potentially just a few lines of code so might be worth thinking about it s probably just a matter of changing these functions in wsgi py def wait on children self while self running try pid status os wait if os wifexited status or os wifsignaled status self remove children pid self verify and respawn children pid status except oserror as err if err errno not in errno eintr errno echild raise except keyboardinterrupt log info li caught keyboard interrupt exiting break except exception sighupinterrupt self reload continue eventlet greenio shutdown safe self sock self sock close log debug exited def run child self def child hup args shuts down child processes existing requests are handled signal signal signal sighup signal sig ign eventlet wsgi is accepting false self sock close pid os fork if pid signal signal signal sighup child hup signal signal signal sigterm signal sig dfl to look like this def wait on children self while self running try pid status os wait if os wifexited status or os wifsignaled status self remove children pid self verify and respawn children pid status except oserror as err if err errno not in errno eintr errno echild raise except keyboardinterrupt log info li caught keyboard interrupt exiting break except exception sighupinterrupt self reload continue except exception sigstopinterrupt self graceful shutdown continue eventlet greenio shutdown safe self sock self sock close log debug exited def run child self def child hup args shuts down child processes existing requests are handled signal signal signal sighup signal sig ign eventlet wsgi is accepting false self sock close def child stop args shuts down child processes existing requests are handled signal signal signal sigstop signal sig ign eventlet wsgi is accepting false self sock close pid os fork if pid signal signal signal sighup child hup signal signal signal sigstop child stop signal signal signal sigterm signal sig dfl and add this one line function def graceful shutdown self send signal to child processes to handle current requests then exit os killpg self pgid signal sigstop i have not tested the above i think it might be sufficient because the existing code will do the accounting of the number of children and exit when self children hits zero but it would need testing you d do kill s stop glance parent process pid	[ 0 ]
here we should remove the if and replace it with dcheck r dest is bit r dest as bitreg r dest	[ 0 ]
this spacing is fucked up	[ 1 ]
I wasn't sure about that variable name, this may bite you in the ass.	[ 1 ]
stop talking implemenation	[ 0 ]
neither of these imports are used anymore they should be removed	[ 0 ]
seems your editor also changed this so another whitespace nit	[ 0 ]
i am slightly concerned that placing two params on the same line violates puppet style did you check this against puppet lint	[ 0 ]
this is nonsense just say ldap also below just use the later expression	[ 0 ]
and volumes ceph images ceph and ephemeral ceph are all false	[ 0 ]
This issue along with not being able to pull manual console changes is just ridiculous. This project could have been something great.	[ 1 ]
Who knows. It\u2019s ridiculous that they have a new `browser_style` boolean property that does almost nothing.	[ 1 ]
nit these two temporaries do not add much value	[ 0 ]
whitespaces kill them all	[ 0 ]
a possibly dumb question will not this command remove also the basic flow entry for allowing the ovs instance to behave as a regular mac based forwarding switch	[ 0 ]
need at least dumb tests for other methods as well	[ 0 ]
given that using memory cache here will have performance and reliability concerns i think a better option would be keeping a cache of all scripts tagged as ads in a separate cache and look that up here	[ 0 ]
these are not the same thing afaict this is introducing a second level of approval for blueprints first you have to get the spec approved at which point you can start working on it then you have to wait for a runway slot before reviewers are permitted to approve it if two reviewers found time to look at a proposed patch have d it the idea that we should block merge due to not having a runway slot is just stupid	[ 0 ]
how will this be assigned the exception was thrown before this occurs in all cases maybe a spacing error	[ 0 ]
instead of wrap delimiter utils wrap delimiter should be used for all wrap delimiter calls below they should be replaced by utils wrap delimiter	[ 0 ]
but you need a way to indicate whether the data is valid or not and that requires either a returning a bool and passing in a pointer since passing a non const reference is against google style or returning a copy and passing in a pointer to a bool something which was explicitly rejected in a much earlier code review i do not honestly know of a third option	[ 0 ]
Fair. Mostly I just wanted to include a representative from each other Hell, but maybe the Tyrant is okay with having an non-representative menagerie.	[ 1 ]
> a antique what the fuck steelpoint	[ 1 ]
The cat GIF is creepy this time...	[ 1 ]
i think this also wants to be const std string	[ 0 ]
dumb question why does the conductor have to do an rpc call to itself	[ 0 ]
you re going to have to rebase i think this whole method gets deleted	[ 0 ]
this might be a dumb question but should not there be a check for physical network to be in self flat networks	[ 0 ]
Fucking cursed brother. Is this even used or am I blind	[ 1 ]
Wearing a poncho over a climate-controlled metal carapace that covers your entire body sounds a bit absurd, though admittedly very stylish. If backpacks are out for "realism" reasons, what about messenger packs, duffel bags, strapped guns and other things you sling over a shoulder? I think fanny packs and other belt items sound like fair game, too. If the oversized pauldrons prevent you from touching your hands together behind your back, you could lay it flat on a table or something and pick its ends up.	[ 1 ]
> Metabolic Interchange is already king of power supply, so we're basically giving the players even fewer reasons to NOT use MI. And it's also utterly ridiculous, you may be noticing a pattern.	[ 1 ]
ed current this spec is based on poc patch c on the poc we are generating swagger spec from the code like validation schema wsgi doc wsgi response as you see on the poc we need to put many decorators on a single api method and generate swagger spec file swagger json the decorators seems redundant and we discussed how about putting a single new decorator swagger or something for clean code for passing swagger definition to api methods	[ 0 ]
The Scala syntax is clumsy to who exactly? It's ND4Scala not ND4(Scala Developers who wish Scala was Numpy).	[ 1 ]
nit alignment should be cleaned up here	[ 0 ]
rst improvements on run the tests using run tests sh n test sql upgrade warning your database will be wiped	[ 0 ]
i propose to change the sentence to if you do not want to install the metapackage you can run the following script to enable remote root access	[ 0 ]
thanks! damn habits...	[ 1 ]
lazy fingers vote for db context	[ 0 ]
anyway either: - change to min 0 so its 50% glass return - ignore it since nobodys gonna make a glass farm wtf	[ 1 ]
took me a minute to realise what the scheduler driver is an eg filterscheduler might be nice for the slow of brain	[ 0 ]
This file and the buildchecker.csproj just seemingly changed. I wasnt sure if everything would fuck itself if i changed it. Should i just get rid of those 2.	[ 1 ]
die is a good function to use here	[ 0 ]
heh well just another thing i will have to sed to python on my machine dumb archlinux	[ 0 ]
Oh, damn. That was my bug.	[ 1 ]
i do not see any profit from that fate sharing with approach implemented in rootwrap daemon you could easily kill daemon without affecting main process and it will get restarted and reload it is configs and code why do you want to get rid of this feature	[ 0 ]
this sucks for ethereals, cant you make the adrenaline stuff a component?	[ 1 ]
this is redundant as the baseadminnetworktest already creates an admin manager so it would be enough to say cls identity manager cls os admin identity client	[ 0 ]
I know. just weird to commit it everytime. It's similar to changing the default value of an unused bool member variable everytime you commit a file lol.	[ 1 ]
sorry for my ignorance but how is bloading bnoui equivalent to busesystemlock isolockfileused	[ 0 ]
is it possible that the current network will not have a key that matches if so then indexerror	[ 0 ]
This was a bitch. I was finally able to implement a working solution; although, I don't feel I completely understand it, at least very well.	[ 1 ]
i think the dcheck line make this if useless	[ 0 ]
shiny	[ 0 ]
done nfs py and glusterfs py use sparsed by the way	[ 0 ]
There's an excess " that fucks everything up ""borg_upgrade_expandedsynthesiser"	[ 1 ]

this is odd why do you need the why not just env exceptionoccurred or just ifs null is there a case where env exceptionoccurred and ifs null if that is the case you will need to clear the exception before you proceed

[ 0 ]

the formatting looks off please run presubmit sh before submitting

[ 0 ]

nit put this below after private

[ 0 ]

if you are trying to minimize wasted work why waste time on basename getprogname

[ 0 ]

why are these lists not assoc but instead this weird struct-esque structure?

[ 1 ]

Damn I tested the example but didn't copy paste it so didn't catch the space.

[ 1 ]

it s effectively dead code dating from before my redesign to create dashboards from json

[ 0 ]

This looks like a crappy test.

[ 1 ]

This was probably a bad interaction of Olympic network tx spamming and the ridiculous I/O load that we threw at leveldb due to a bug in state DB dirty tracking.

[ 1 ]

fall through the end with self neutron service plugins none which was the original behaviour

[ 0 ]

i cannot quite tell but does not this fit without the escape char port dict self core plugin make port dict port process extensions false

[ 0 ]

> realistic use cases for these things require making copies of that buffer, and then you're screwed again Why are you screwed? As long as you `Py_INCREF` after making that copy and before doing anything with it, you're fine, right? > I did a quick skim of numpy and scipy and every use looks broken to me. *When combined with this patch*, numpy looks alright to me, in the one place we use it. That doesn't mean this is a good api, but it does look like a usable one.

[ 1 ]

Damn. This crash never showed up in debug mode for some reason but only in release build. Fix [1957eba]

[ 1 ]

hmm returning here unconditionally seems wrong we have not run validatejsonstring for the correctly typed entries after fixing this it would be good to have a unit test for this case

[ 0 ]

My trend is always to avoid Net::HTTP (it's slow, clumsy, and buggy) I use ftw (I wrote it) instead of Net::HTTP because that library is quite horrible Open to accepting patches to add proxy support to ftw ;)

[ 1 ]

should this also pass through the authorization code ttl

[ 0 ]

Hopefully it's better now, but I suck at complex sentences.

[ 1 ]

This sounds very ridiculous, I'm not sure if we really want it?

[ 1 ]

did you try /atom/Click or /atom/proc/Click? using . here would i think fuck you up

[ 1 ]

could use a list of turfs instead of world, pretty much anything "in world" is shit for lag.

[ 1 ]

how about doing this when dealing with the stringfactory you would then only push new instance of strings

[ 0 ]

either const iterator findintervalstart uint t from or std pair bool int t find uint t from the from and to are also bad names but let us not change them in this cl

[ 0 ]

only a stub info are in the wadl file

[ 0 ]

please throw a nullpointerexception

[ 0 ]

realized essex should be removed

[ 0 ]

U fuckin faggot I swear to God I'll find u

[ 1 ]

you could replace your whole logic with a simple for loop size t i for i i count use buf i

[ 0 ]

yes none or insignificant if you prefer importlib or not you need to call shell commands to plug your interface like ovs vsctl in a shell

[ 0 ]

Because breaking functions is a fucking good idea. Can't you keep it on del process and not use qdel?

[ 1 ]

nit false is scan request

[ 0 ]

what do you plan for deprecating the old value in order to avoid confusion for people wanting to update retries count

[ 0 ]

it works in trunk and diablo stable is there a reason you are testing against the rcb build i am not aware of that being a publicly available build or that we need to be testing against it

[ 0 ]

what are you take into account in order to simulation a migration live block dirty page bandwidth

[ 0 ]

get rid of bits in byte you can do div round up add out pixelbits instead

[ 0 ]

Colour is great - thanks I am not a css expert but I am pretty sure that the css can be written in shorthand instead of all those classes being specified directly. @C-Lodder @ciar4n am I correct or talking the usual rubbish

[ 1 ]

the sprite is entirely blob colors, so it'd be a bit weird having dough in the recipe but no dough in the sprite

[ 1 ]

let s eradicate the black magic stdout stderr utils execute chroot chroot dpkg query w f package version pkgver list stdout split

[ 0 ]

In GitLab by @jeremyletang on Nov 21, 2019, 14:29 yes it's nonsense indeed.

[ 1 ]

scratch that i was being an idiot

[ 0 ]

you should absolutely remove the period here plog adds a colon and an error string

[ 0 ]

manager should be removed here

[ 0 ]

makes sense but if that is the kind of restriction someone wants to enforce could not they just use the policies setting instead of the roles i do not mean to be a jerk i am just trying to understand what extending the semantics of the roles style property protections gets you that is not already available from the policies style property protections

[ 0 ]

stupid nit added vs available or accessible from the client perspective maybe

[ 0 ]

You can just call animate() with a 3 like that???!!? wtf byond pls Even if you can, I would probably make it time = 3 just for clarity

[ 1 ]

remove the ifdef here

[ 0 ]

ah i would forgotten about the disabled by file part of the middlware i guess the difference between a graceful shutdown and using the middleware is that one is event driven where the other one requires you to pick a long time and assume that after that there are no requests being handled which would work but the code to implement the graceful shutdown could be pretty easy potentially just a few lines of code so might be worth thinking about it s probably just a matter of changing these functions in wsgi py def wait on children self while self running try pid status os wait if os wifexited status or os wifsignaled status self remove children pid self verify and respawn children pid status except oserror as err if err errno not in errno eintr errno echild raise except keyboardinterrupt log info li caught keyboard interrupt exiting break except exception sighupinterrupt self reload continue eventlet greenio shutdown safe self sock self sock close log debug exited def run child self def child hup args shuts down child processes existing requests are handled signal signal signal sighup signal sig ign eventlet wsgi is accepting false self sock close pid os fork if pid signal signal signal sighup child hup signal signal signal sigterm signal sig dfl to look like this def wait on children self while self running try pid status os wait if os wifexited status or os wifsignaled status self remove children pid self verify and respawn children pid status except oserror as err if err errno not in errno eintr errno echild raise except keyboardinterrupt log info li caught keyboard interrupt exiting break except exception sighupinterrupt self reload continue except exception sigstopinterrupt self graceful shutdown continue eventlet greenio shutdown safe self sock self sock close log debug exited def run child self def child hup args shuts down child processes existing requests are handled signal signal signal sighup signal sig ign eventlet wsgi is accepting false self sock close def child stop args shuts down child processes existing requests are handled signal signal signal sigstop signal sig ign eventlet wsgi is accepting false self sock close pid os fork if pid signal signal signal sighup child hup signal signal signal sigstop child stop signal signal signal sigterm signal sig dfl and add this one line function def graceful shutdown self send signal to child processes to handle current requests then exit os killpg self pgid signal sigstop i have not tested the above i think it might be sufficient because the existing code will do the accounting of the number of children and exit when self children hits zero but it would need testing you d do kill s stop glance parent process pid

[ 0 ]

here we should remove the if and replace it with dcheck r dest is bit r dest as bitreg r dest

[ 0 ]

this spacing is fucked up

[ 1 ]

I wasn't sure about that variable name, this may bite you in the ass.

[ 1 ]

stop talking implemenation

[ 0 ]

neither of these imports are used anymore they should be removed

[ 0 ]

seems your editor also changed this so another whitespace nit

[ 0 ]

i am slightly concerned that placing two params on the same line violates puppet style did you check this against puppet lint

[ 0 ]

this is nonsense just say ldap also below just use the later expression

[ 0 ]

and volumes ceph images ceph and ephemeral ceph are all false

[ 0 ]

This issue along with not being able to pull manual console changes is just ridiculous. This project could have been something great.

[ 1 ]

Who knows. It\u2019s ridiculous that they have a new `browser_style` boolean property that does almost nothing.

[ 1 ]

nit these two temporaries do not add much value

[ 0 ]

whitespaces kill them all

[ 0 ]

a possibly dumb question will not this command remove also the basic flow entry for allowing the ovs instance to behave as a regular mac based forwarding switch

[ 0 ]

need at least dumb tests for other methods as well

[ 0 ]

given that using memory cache here will have performance and reliability concerns i think a better option would be keeping a cache of all scripts tagged as ads in a separate cache and look that up here

[ 0 ]

these are not the same thing afaict this is introducing a second level of approval for blueprints first you have to get the spec approved at which point you can start working on it then you have to wait for a runway slot before reviewers are permitted to approve it if two reviewers found time to look at a proposed patch have d it the idea that we should block merge due to not having a runway slot is just stupid

[ 0 ]

how will this be assigned the exception was thrown before this occurs in all cases maybe a spacing error

[ 0 ]

instead of wrap delimiter utils wrap delimiter should be used for all wrap delimiter calls below they should be replaced by utils wrap delimiter

[ 0 ]

but you need a way to indicate whether the data is valid or not and that requires either a returning a bool and passing in a pointer since passing a non const reference is against google style or returning a copy and passing in a pointer to a bool something which was explicitly rejected in a much earlier code review i do not honestly know of a third option

[ 0 ]

Fair. Mostly I just wanted to include a representative from each other Hell, but maybe the Tyrant is okay with having an non-representative menagerie.

[ 1 ]

> a antique what the fuck steelpoint

[ 1 ]

The cat GIF is creepy this time...

[ 1 ]

i think this also wants to be const std string

[ 0 ]

dumb question why does the conductor have to do an rpc call to itself

[ 0 ]

you re going to have to rebase i think this whole method gets deleted

[ 0 ]

this might be a dumb question but should not there be a check for physical network to be in self flat networks

[ 0 ]

Fucking cursed brother. Is this even used or am I blind

[ 1 ]

Wearing a poncho over a climate-controlled metal carapace that covers your entire body sounds a bit absurd, though admittedly very stylish. If backpacks are out for "realism" reasons, what about messenger packs, duffel bags, strapped guns and other things you sling over a shoulder? I think fanny packs and other belt items sound like fair game, too. If the oversized pauldrons prevent you from touching your hands together behind your back, you could lay it flat on a table or something and pick its ends up.

[ 1 ]

> Metabolic Interchange is already king of power supply, so we're basically giving the players even fewer reasons to NOT use MI. And it's also utterly ridiculous, you may be noticing a pattern.

[ 1 ]

ed current this spec is based on poc patch c on the poc we are generating swagger spec from the code like validation schema wsgi doc wsgi response as you see on the poc we need to put many decorators on a single api method and generate swagger spec file swagger json the decorators seems redundant and we discussed how about putting a single new decorator swagger or something for clean code for passing swagger definition to api methods

[ 0 ]

The Scala syntax is clumsy to who exactly? It's ND4Scala not ND4(Scala Developers who wish Scala was Numpy).

[ 1 ]

nit alignment should be cleaned up here

[ 0 ]

rst improvements on run the tests using run tests sh n test sql upgrade warning your database will be wiped

[ 0 ]

i propose to change the sentence to if you do not want to install the metapackage you can run the following script to enable remote root access

[ 0 ]

thanks! damn habits...

[ 1 ]

lazy fingers vote for db context

[ 0 ]

anyway either: - change to min 0 so its 50% glass return - ignore it since nobodys gonna make a glass farm wtf

[ 1 ]

took me a minute to realise what the scheduler driver is an eg filterscheduler might be nice for the slow of brain

[ 0 ]

This file and the buildchecker.csproj just seemingly changed. I wasnt sure if everything would fuck itself if i changed it. Should i just get rid of those 2.

[ 1 ]

die is a good function to use here

[ 0 ]

heh well just another thing i will have to sed to python on my machine dumb archlinux

[ 0 ]

Oh, damn. That was my bug.

[ 1 ]

i do not see any profit from that fate sharing with approach implemented in rootwrap daemon you could easily kill daemon without affecting main process and it will get restarted and reload it is configs and code why do you want to get rid of this feature

[ 0 ]

this sucks for ethereals, cant you make the adrenaline stuff a component?

[ 1 ]

this is redundant as the baseadminnetworktest already creates an admin manager so it would be enough to say cls identity manager cls os admin identity client

[ 0 ]

I know. just weird to commit it everytime. It's similar to changing the default value of an unused bool member variable everytime you commit a file lol.

[ 1 ]

sorry for my ignorance but how is bloading bnoui equivalent to busesystemlock isolockfileused

[ 0 ]

is it possible that the current network will not have a key that matches if so then indexerror

[ 0 ]

This was a bitch. I was finally able to implement a working solution; although, I don't feel I completely understand it, at least very well.

[ 1 ]

i think the dcheck line make this if useless

[ 0 ]

shiny

[ 0 ]

done nfs py and glusterfs py use sparsed by the way

[ 0 ]

There's an excess " that fucks everything up ""borg_upgrade_expandedsynthesiser"

[ 1 ]