@m.shulhan @hervinhioslash
Thanks for the suggestion. I agree that using a runtime mechanism like env vars
or command line flag can solve this problem. But the path value is for internal
use, it should be customized only by the system administrators or package
managers who determine where to
> Whats wrong with const?
Const is the ideal choice, but how do users of my program specify the path when
they compile it?
> And could not be global var then?
Using var loses the guarantee that the path won’t be changed, and the go
compiler can no longer optimize as much I presume?
--
You
Thanks, as mentioned in the previous reply, this is what I end up using.
On Thursday, April 16, 2020 at 10:25:03 AM UTC+8, Shulhan wrote:
>
>
> > On 15 Apr 2020, at 18.55, Glen Huang >
> wrote:
> >
> > I have a makefile where upon a new go binary being built,
e go build directly.
move-if-change sounds like a good approach, will come in handy when there
is no need to cross compile, thanks for the tip.
On Thursday, April 16, 2020 at 4:09:27 AM UTC+8, Ian Lance Taylor wrote:
>
> On Wed, Apr 15, 2020 at 4:55 AM Glen Huang >
> wrote:
> >
>
I have a makefile where upon a new go binary being built, it builds some
other stuff that is not go related.
The go binary is unconditionally built with a FORCE prerequisite, so the go
build command always runs, but that command always updates the output
binary, which leads to downstream being
@lan @Shulhan
Thanks. Glad to know the design goal much better.
On Tuesday, April 14, 2020 at 1:07:36 PM UTC+8, Ian Lance Taylor wrote:
>
> On Mon, Apr 13, 2020 at 6:56 PM Glen Huang >
> wrote:
> >
> > Given
> >
> > type Data []byte
> > s :
on of this type, since one structure isn't
> safely coercible to the other type. This would be along the lines of
> reinterpret_cast<> in c++, which can corrupt memory.
>
>
> On Mon, Apr 13, 2020 at 6:56 PM Glen Huang >
> wrote:
>
>> Given
>>
Given
type Data []byte
s := [][]byte{{1},{2}}
I wonder why this conversion isn't allowed?
[]Data(s)
Seems pretty straightforward. Maybe I miss some edge cases where things
won't square?
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To
ible.
>
> On Sun, Feb 23, 2020 at 10:24 AM Glen Huang > wrote:
>
>> Another slightly related topic:
>>
>> switch v := i.(type) {
>> case map[string]interface{}, map[string]int:
>> fmt.Print(v)
>> }
>>
>> Ideally v is of type map[strin
Another slightly related topic:
switch v := i.(type) {
case map[string]interface{}, map[string]int:
fmt.Print(v)
}
Ideally v is of type map[string]interface{}, but it's interface{} instead.
On Sunday, February 23, 2020 at 5:12:27 PM UTC+8, Glen Huang wrote:
>
> Hi,
>
> I hav
Hi,
I have a function that accepts an argument of type map[string]interface{},
and I also have a value of type map[string]int.
Currently it seems I can't directly pass the value to the function. Is
there anyway I can directly coerce it or a new value of the exact matching
type must be
you "inherit" these methods and you don't want this to happe
Given the case above, I'm wondering if embedding + not overriding method is
possible?
On Thursday, January 2, 2020 at 9:19:23 AM UTC+8, Ben Bullock wrote:
>
>
>
> On Tuesday, 31 December 2019 19:21:07 UTC+9, Glen
After thinking more about it, I start to feel that maybe using embedded
structs is generally a bad idea with dealing with json? Is the correct
solution here to use explicit fields?
BTW, happy New Year everyone.
On Tuesday, December 31, 2019 at 6:21:07 PM UTC+8, Glen Huang wrote:
>
>
I want to unmarshal a struct that contains embedded structs:
type Parent struct {
Child
P int
}
type Child struct {
Grandchild
C int
}
type Grandchild struct {
G int
}
The problem is that Grandchild defines its own UnmarshalJSON method, and
that means Parent inherits it and breaks
on hiatus) right now?
On Monday, March 25, 2019 at 10:36:12 PM UTC+8, Sameer Ajmani wrote:
>
> With gRPC, I recommend you use protobuf.
>
> On Mon, Mar 25, 2019 at 8:18 AM Glen Huang >
> wrote:
>
>> I planed to use net/rpc with gob, but then found the GitHub issue sayin
I planed to use net/rpc with gob, but then found the GitHub issue saying
net/rpc is deprecated, and people should be using grpc instead.
That leads to the question should I use grpc with gob or stick with
protobuf? Google suggests not many people talk about using grpc with gob,
any there
So looks like nftables is my only choice then?
On Friday, March 15, 2019 at 10:30:28 PM UTC+8, Andrei Tudor Călin wrote:
>
> That sounds like something a firewall would do, not your Go program.
>
> On 3/15/19 3:14 PM, Glen Huang wrote:
> > Thanks, but if I'm not wrong, that m
>
> Then, to use:
>
> ln, err := net.Listen("tcp", addr)
> if err != nil {
> log.Fatal(err)
> }
> aln := {allowed: yourListOfIPs, inner: ln}
> tlsln := tls.NewListener(aln, yourTLSConfig)
>
> // use tlsln
>
> On 3/15/19 2:58 PM,
checks the list of allowed
> IPs.
> You'll be able to run code before the connection is passed on to
> crypto/tls.
> Wrap it using https://golang.org/pkg/crypto/tls/#NewListener.
>
> On 3/15/19 2:10 PM, Glen Huang wrote:
> > I'm trying to limit which clients are allowed
I'm trying to limit which clients are allowed to connect to my tls server
by their IPs.
I know I can do that after Accept, check their IPs and close the connection
if they're not whitelisted. But that means the full tls handshake has to
complete before I can do that.
Another option is that I
Thanks for the reply.
Hopefully the doc can be changed to avoid confusion.
On Sunday, January 21, 2018 at 1:16:04 PM UTC+8, Ian Lance Taylor wrote:
>
> On Sat, Jan 20, 2018 at 8:43 PM, Glen Huang <hey...@gmail.com
> > wrote:
> >
> > The doc for the method
Hi,
The doc for the method type in the reflect package
(https://golang.org/pkg/reflect/#Method) seems to suggest that it's
possible to inspect a type's unexported methods:
// PkgPath is the package path that qualifies a lower case
(unexported)
// method name. It is empty for
ervice
> }
>
>
> func(h handleLogin)ServeHTTP(w http.ResponseWriter,r *http.Request){
> h.Service.CallThis(r)
> }
>
> You can only write unit tests if you write testable code at first place.
>
> Le mercredi 15 novembre 2017 03:28:28 UTC+1, Glen Huang a écrit :
>>
&g
I find that to be able to unit test things in golang, I need to restructure
them in an uncomfortable way in order to be able to mock their dependencies.
For example, say I want to test this simple http handler:
func handleLogIn(w http.ResponseWriter, r *http.Request) {
svc :=
Any suggestion how to improve compressor and resizer's APIs?
Regards,
Glen
On Thursday, November 9, 2017 at 12:26:14 PM UTC+8, Glen Huang wrote:
>
> I'm writing a small image processing tool, given an image content, it only
> does two things:
>
> 1. compressing the image content
I'm writing a small image processing tool, given an image content, it only
does two things:
1. compressing the image content
2. generating a thumbnail (by resizing and then compressing it)
I want to model the compressing(the same goes for resizing) after the
compress stdlib: you create a
ave succeeded, or failed for another reason, before the
> context was cancelled. Whether this matters is of course up to the
> application.
>
> //jb
>
>
>> On 8 Nov 2017, at 13:19, Jan Mercl <0xj...@gmail.com
>> <mailto:0xj...@gmail.com>> wrote:
>&
> switch;
>
> _, err := http.DefaultClient.Do(r)
> if urlErr, ok := err.(*url.Error); ok {
> log.Println(urlErr.Err == context.Canceled) // “true", in your example
> }
>
> //jb
>
>
> On 8 Nov 2017, at 13:11, Glen Huang <hey...@gmail.com >
> wrote:
>
>
mber 2017 14:02.13 UTC+2 kirjutas Glen Huang:
>>
>> I have this simple code in which I try to check if the request was
>> cancelled. But surprisingly, it prints false instead of true in go 1.9.
>>
>> I wonder what's the correct way to check that?
>>
>> pa
Jul 25, 2017 at 6:56 AM Glen Huang <hey...@gmail.com >
> wrote:
>
>>
>> My current design is that uploading is finished as soon as the image is
>> uploaded to the server, after which I spawn a goroutine to exec a command
>> to do the compression. Wh
oh, never thought about that. Initially there will be only server but it's
only matter of time I need to scale to multiple servers. I'll probably give
redis a shot.
Thanks for the suggestion!
On Wednesday, July 26, 2017 at 8:33:59 AM UTC+8, Florin Pățan wrote:
>
> Will the user always end up
Thanks for bringing singleflight to my attention.
However, if I'm not wrong, it doesn't seem to fully solve the problem.
If after uploading, I call singleflight.Do, and pass the file name and the
compression function, when users requests a file, should I call
singleflight.Do and pass the
lable
> by searching for those keywords.
>
> On Tuesday, 25 July 2017 01:34:30 UTC+10, Glen Huang wrote:
>>
>> Hi,
>>
>> I'm still pretty new to go. Hope this question isn't too stupid.
>>
>> I'm writing a restful API server, and in order to send a respon
ill cause issues, and even if they do, solutions
> such as adding indexes to problematic fields or doing caching might help A
> LOT more than trying to be more concurrent will.
>
>
> On Monday, July 24, 2017 at 8:34:30 AM UTC-7, Glen Huang wrote:
>>
>> Hi,
>>
>&g
Hi,
I'm still pretty new to go. Hope this question isn't too stupid.
I'm writing a restful API server, and in order to send a response, I need
to query a db to get its content type and then send the actually file that
lives on the file system. Now the question is, should I put db.QueryRow and
35 matches
Mail list logo
