August 2024 - Wine-GitLab - list.winehq.org

[PATCH 0/4] MR6216: Implement httpapi HttpResponseSendEntityBody
by Tom Helander (＠thelande) Aug. 6, 2024

Aug. 6, 2024

This merge request implements the `httpapi.dll` `HttpResponseSendEntityBody` function and associated tests. API documentation was written by me by paraphrasing/re-writing what I had read for the function from the MSDN documentation. I am adding this function to enable the remote API component of the Space Engineers Dedicated Server. Implementation notes: - I mainly copied the relevant portion of `HttpSendHttpResponse` for `HttpResponseSendEntityBody`. - I updated the `http_response` struct to add a field `response_flags` - This field tracks the flags passed in to `HttpSendHttpResponse` and `HttpResponseSendEntityBody`. - This was needed because the clean-up steps in `http.sys` need to be skipped when `HTTP_SEND_RESPONSE_FLAG_MORE_DATA` is set. - I updated `HttpSendHttpResponse` to handle the `HTTP_SEND_RESPONSE_FLAG_MORE_DATA` flag by omitting the `Content-Length` header and setting the `response_flags` field in the `response` object passed to `http.sys`. - The `Content-Length` header is omitted by the Windows implementation and thus I omitted it here for conformance purposes. - I updated `http_send_response` in `http.sys` to skip the connection clean up steps if the `HTTP_SEND_RESPONSE_FLAG_MORE_DATA` flag was set. This works for both `HttpSendHttpResponse` and `HttpResponseSendEntityBody` as either could set the flag to continue to send more data. - It is not 100% clear to me what the cleanup steps are doing. In particular, I am uncertain what lines 1011-1016 and 1023-1027 are doing, but based on context it seemed appropriate to keep them with the other connection cleanup steps. For testing, I duplicated the existing `HttpSendHttpResponse` test but modified it to set the `HTTP_SEND_RESPONSE_FLAG_MORE_DATA` flag and then followed up with a call to `HttpResponseSendEntityBody` with new data. The validation ensures that data from both the initial response and follow up response are contained in the final response obtained via `recv()`. On Windows 11 this happens immediately without any delay, but in Wine I had to add a `Sleep(100)` to give enough time for the follow up response to be received in the socket buffer by the time the `recv()` call was made in the test. I could have looped the `recv()` call to collect all of the response, but that felt more complicated than adding a minor delay. The tests were verified against Windows 11 Pro (build 22621.3880). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/6216

3 5

[PATCH v2 0/7] MR6212: mshtml: Use proper JS functions for constructors.
by Jacek Caban (＠jacek) Aug. 6, 2024

Aug. 6, 2024

-- v2: mshtml: Add support for MutationObserver consturctor and prototype objects. mshtml: Use host object script bindings for Option constructor object. mshtml: Add support for option element prototype objects. mshtml: Use host constructor script bindings for XMLHttpRequest constructor object. mshtml: Use host constructor script bindings for Image constructor object. jscript: Introduce HostConstructor function type. https://gitlab.winehq.org/wine/wine/-/merge_requests/6212

3 8

[PATCH 0/1] MR6215: ntdll: Try to use page size from host_page_size() for macOS.
by Jactry Zeng (＠jactry) Aug. 6, 2024

Aug. 6, 2024

For macOS on Apple Silion, includes Rosetta 2, page size is 16K by default. This fixes the incorrect ullAvailPageFile returned by GlobalMemoryStatusEx(). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/6215

3 2

[PATCH v4 0/1] MR6191: ws2_32: Fully implement socket options IPX_ADDRESS and IPX_MAX_ADAPTER_NUM on Linux
by Haoran Wang (＠Low-power) Aug. 6, 2024

Aug. 6, 2024

Tested using C&C Red Alert 2; selection of interface and network communication worked as expected. IPX traffic correctly sent to the selected interface only, as confirmed by **tcpdump(8)**. ![ra2-on-linux-with-multiple-ipx-interfaces](/uploads/7755db6a7b82a21050623d3ccd9ec37e/ra2-on-linux-with-multiple-ipx-interfaces.png) -- v4: ws2_32: Fully implement socket options IPX_ADDRESS and IPX_MAX_ADAPTER_NUM on Linux https://gitlab.winehq.org/wine/wine/-/merge_requests/6191

2 1

[PATCH v2 0/1] MR6191: ws2_32: Fully implement socket options IPX_ADDRESS and IPX_MAX_ADAPTER_NUM on Linux
by Haoran Wang (＠Low-power) Aug. 6, 2024

Aug. 6, 2024

Tested using C&C Red Alert 2; selection of interface and network communication worked as expected. IPX traffic correctly sent to the selected interface only, as confirmed by **tcpdump(8)**. ![ra2-on-linux-with-multiple-ipx-interfaces](/uploads/7755db6a7b82a21050623d3ccd9ec37e/ra2-on-linux-with-multiple-ipx-interfaces.png) -- v2: ws2_32: Fully implement socket options IPX_ADDRESS and IPX_MAX_ADAPTER_NUM on Linux https://gitlab.winehq.org/wine/wine/-/merge_requests/6191

2 2

[PATCH 0/5] MR6214: hidclass: Use a single device reading thread instead of one for each TLC.
by Rémi Bernon Aug. 6, 2024

Aug. 6, 2024

Otherwise, with multi-TLC devices, concurrent PDO threads would steal reports from each other, which causes random timeouts in the tests. Fixes: c65d5b094825afdafba462a4050a490426f3cd51 -- https://gitlab.winehq.org/wine/wine/-/merge_requests/6214

1 5

Re: [PATCH v13 0/1] MR6150: win32u: fix a regression bug that virtual desktop not effective.
by Rémi Bernon Aug. 6, 2024

Aug. 6, 2024

I have opened https://gitlab.winehq.org/wine/wine/-/merge_requests/6213 which I believe should fix the issue in a simpler way. Note that it's not completely correct in any case, and there might be some unexpected behavior if applications rely on monitor display settings. We still allow different virtual desktop window sizes while the underlying virtual desktop size should always be the same as the monitor rectangles, which is normally (on Windows) the same across all the desktops of the interactive window station. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/6150#note_77860

1 0

[PATCH 0/1] MR6213: win32u: Force update the display cache when virtual desktop state changes.
by Rémi Bernon Aug. 6, 2024

Aug. 6, 2024

This can happen when explorer creates a new virtual desktop, while being associated with the default non-virtual desktop. Explorer then fails to change the display settings of the virtual source in this case. Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=56993 -- https://gitlab.winehq.org/wine/wine/-/merge_requests/6213

1 1

[PATCH 0/1] MR6191: ws2_32: Fully implement socket options IPX_ADDRESS and IPX_MAX_ADAPTER_NUM on Linux
by Haoran Wang (＠Low-power) Aug. 6, 2024

Aug. 6, 2024

Move implementation of these 2 options into ntdll, and add 2 ioctl codes for calling from ws2_32. Query configured IPX interfaces by reading /proc/net/ipx/interface of Linux procfs, enabling applications to iterate multiple IPX interfaces. This can be useful when a node have multiple IPX interfaces, and application need to receive/send packets from/to a non-primary IPX interface. Signed-off-by: WHR <whr(a)rivoreo.one> -- https://gitlab.winehq.org/wine/wine/-/merge_requests/6191

4 9

[PATCH 0/7] MR6212: mshtml: Use proper JS functions for constructors.
by Jacek Caban (＠jacek) Aug. 6, 2024

Aug. 6, 2024

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/6212

3 9