Smalltalk › Squeak › Squeak - Dev

browseMcMethodRevisions Request-URI Too Large

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

5 messages Options

Nicolas Cellier

browseMcMethodRevisions Request-URI Too Large

Hi All,

when trying to browse revisions for some long methods (some VMMaker methods are long) this often lead to failure Notifier.

Indeed, it seems like {self serializeForRequest: aMCDefinition} is serializing the whole method source...

You might want to try with SistaRegisterAllocatingCogit>>genSpecialSelectorComparison

Whole reply is:

HTTP/1.1 414 Request-URI Too Large
server: nginx/1.16.0
date: Mon, 07 Oct 2019 20:55:41 GMT
content-type: text/html
content-length: 177
connection: close

<html>
<head><title>414 Request-URI Too Large</title></head>
<body>
<center><h1>414 Request-URI Too Large</h1></center>
<hr><center>nginx/1.16.0</center>
</body>
</html>

Is this (URI length) something we can tune at server side?

Levente Uzonyi

Re: browseMcMethodRevisions Request-URI Too Large

On Mon, 7 Oct 2019, Nicolas Cellier wrote:

> Hi All,
> when trying to browse revisions for some long methods (some VMMaker methods are long) this often lead to failure Notifier.
>
> Indeed, it seems like {self serializeForRequest: aMCDefinition} is serializing the whole method source...
>
> You might want to try with SistaRegisterAllocatingCogit>>genSpecialSelectorComparison
>
> Whole reply is:
>
> HTTP/1.1 414 Request-URI Too Large
> server: nginx/1.16.0
> date: Mon, 07 Oct 2019 20:55:41 GMT
> content-type: text/html
> content-length: 177
> connection: close
>
> <html>
> <head><title>414 Request-URI Too Large</title></head>
> <body>
> <center><h1>414 Request-URI Too Large</h1></center>
> <hr><center>nginx/1.16.0</center>
> </body>
> </html>
>
> Is this (URI length) something we can tune at server side?

Yes, it could be tuned, but that wouldn't solve the problem. URIs should
not be arbitrarily long.
The proper solution would be to make the server expect POST requests with
the content sent in the request body, and the image sending matching POST
requests instead of GET request.

Btw, we had a discussion about this issue, but nothing happened since
then.

Levente

>
>
>

Chris Muller-3

Re: browseMcMethodRevisions Request-URI Too Large

On Mon, Oct 7, 2019 at 4:41 PM Levente Uzonyi <[hidden email]> wrote:

On Mon, 7 Oct 2019, Nicolas Cellier wrote:

> Hi All,
> when trying to browse revisions for some long methods (some VMMaker methods are long) this often lead to failure Notifier.
>
> Indeed, it seems like {self serializeForRequest: aMCDefinition} is serializing the whole method source...
>
> You might want to try with SistaRegisterAllocatingCogit>>genSpecialSelectorComparison
>
> Whole reply is:
>
> HTTP/1.1 414 Request-URI Too Large
> server: nginx/1.16.0
> date: Mon, 07 Oct 2019 20:55:41 GMT
> content-type: text/html
> content-length: 177
> connection: close
>
> <html>
> <head><title>414 Request-URI Too Large</title></head>
> <body>
> <center><h1>414 Request-URI Too Large</h1></center>
> <hr><center>nginx/1.16.0</center>
> </body>
> </html>
>
> Is this (URI length) something we can tune at server side?

Yes, it could be tuned, but that wouldn't solve the problem. URIs should
not be arbitrarily long.
The proper solution would be to make the server expect POST requests with
the content sent in the request body, and the image sending matching POST
requests instead of GET request.

Btw, we had a discussion about this issue, but nothing happened since
then.

It came up a couple of times this year. Some background and the method needing fixed, in case someone has time to do it.

http://lists.squeakfoundation.org/pipermail/squeak-dev/2019-July/202772.html

Levente Uzonyi

Re: browseMcMethodRevisions Request-URI Too Large

Hi Chris,

On Wed, 9 Oct 2019, Chris Muller wrote:

> On Mon, Oct 7, 2019 at 4:41 PM Levente Uzonyi <[hidden email]> wrote:
> On Mon, 7 Oct 2019, Nicolas Cellier wrote:
>
> > Hi All,
> > when trying to browse revisions for some long methods (some VMMaker methods are long) this often lead to failure Notifier.
> >
> > Indeed, it seems like {self serializeForRequest: aMCDefinition} is serializing the whole method source...
> >
> > You might want to try with SistaRegisterAllocatingCogit>>genSpecialSelectorComparison
> >
> > Whole reply is:
> >
> > HTTP/1.1 414 Request-URI Too Large
> > server: nginx/1.16.0
> > date: Mon, 07 Oct 2019 20:55:41 GMT
> > content-type: text/html
> > content-length: 177
> > connection: close
> >
> > <html>
> > <head><title>414 Request-URI Too Large</title></head>
> > <body>
> > <center><h1>414 Request-URI Too Large</h1></center>
> > <hr><center>nginx/1.16.0</center>
> > </body>
> > </html>
> >
> > Is this (URI length) something we can tune at server side?
>
> Yes, it could be tuned, but that wouldn't solve the problem. URIs should
> not be arbitrarily long.
> The proper solution would be to make the server expect POST requests with
> the content sent in the request body, and the image sending matching POST
> requests instead of GET request.
>
> Btw, we had a discussion about this issue, but nothing happened since
> then.
>
>
> It came up a couple of times this year. Some background and the method needing fixed, in case someone has time to do it.
>
> http://lists.squeakfoundation.org/pipermail/squeak-dev/2019-July/202772.html

I doubt mixing PUT and POST requests is a good idea. It would be better to
create a separate POST handler for this.

Levente

>
>
>

Chris Muller-3

Re: browseMcMethodRevisions Request-URI Too Large

Oh my, it appears I conflated PUT and POST, maybe the fix would be easier than I thought then!

This is still far down on my list right now, though. We should have a backup for me, anyway. I nominate Tim. :)

- Chris

On Wed, Oct 9, 2019 at 4:49 PM Levente Uzonyi <[hidden email]> wrote:

Hi Chris,

On Wed, 9 Oct 2019, Chris Muller wrote:

> On Mon, Oct 7, 2019 at 4:41 PM Levente Uzonyi <[hidden email]> wrote:
> On Mon, 7 Oct 2019, Nicolas Cellier wrote:
>
> > Hi All,
> > when trying to browse revisions for some long methods (some VMMaker methods are long) this often lead to failure Notifier.
> >
> > Indeed, it seems like {self serializeForRequest: aMCDefinition} is serializing the whole method source...
> >
> > You might want to try with SistaRegisterAllocatingCogit>>genSpecialSelectorComparison
> >
> > Whole reply is:
> >
> > HTTP/1.1 414 Request-URI Too Large
> > server: nginx/1.16.0
> > date: Mon, 07 Oct 2019 20:55:41 GMT
> > content-type: text/html
> > content-length: 177
> > connection: close
> >
> > <html>
> > <head><title>414 Request-URI Too Large</title></head>
> > <body>
> > <center><h1>414 Request-URI Too Large</h1></center>
> > <hr><center>nginx/1.16.0</center>
> > </body>
> > </html>
> >
> > Is this (URI length) something we can tune at server side?
>
> Yes, it could be tuned, but that wouldn't solve the problem. URIs should
> not be arbitrarily long.
> The proper solution would be to make the server expect POST requests with
> the content sent in the request body, and the image sending matching POST
> requests instead of GET request.
>
> Btw, we had a discussion about this issue, but nothing happened since
> then.
>
>
> It came up a couple of times this year. Some background and the method needing fixed, in case someone has time to do it.
>
> http://lists.squeakfoundation.org/pipermail/squeak-dev/2019-July/202772.html

I doubt mixing PUT and POST requests is a good idea. It would be better to
create a separate POST handler for this.

Levente

>
>
>