Reduce windows executable size by excluding unused libs #1678

2020-10-01T17:38:19+02:00

g1itch commented

2020-10-01 17:38:19 +02:00

(Migrated from github.com)

Hello!

I formatted bitmessagemain.spec and excluded unused libraries. This should slightly reduce the size of windows exe.

Hello! I formatted bitmessagemain.spec and excluded unused libraries. This should slightly reduce the size of windows exe.

PeterSurda (Migrated from github.com) approved these changes 2020-10-05 09:32:05 +02:00

PeterSurda (Migrated from github.com) left a comment

Looks OK, builds succeed. In the future, windows builds will be done automatically for every PR, you can't download the resulting binaries but at least you'll have the build log available.

g1itch commented

2020-10-05 12:40:10 +02:00

(Migrated from github.com)

Looks OK, builds succeed. In the future, windows builds will be done automatically for every PR, you can't download the resulting binaries but at least you'll have the build log available.

It fails to run on Windows 10. So I don't merge, it needs investigation.

> Looks OK, builds succeed. In the future, windows builds will be done automatically for every PR, you can't download the resulting binaries but at least you'll have the build log available. It fails to run on Windows 10. So I don't merge, it needs investigation.

PeterSurda commented

2020-10-05 12:52:06 +02:00

(Migrated from github.com)

64 bit builds for windows 10 aren't working. This is a separate issue and a developer is working on it: #1622

g1itch commented

2020-10-05 13:01:46 +02:00

(Migrated from github.com)

64 bit builds for windows 10 aren't working. This is a separate issue and a developer is working on it: #1622

I read it and tested i386 build.

> 64 bit builds for windows 10 aren't working. This is a separate issue and a developer is working on it: #1622 I read it and tested i386 build.

👍 1

g1itch commented

2020-10-08 11:57:59 +02:00

(Migrated from github.com)

Hmm... Today it runs:

2020-10-08 12:29:23,505 - WARNING - Using default logger configuration
2020-10-08 12:29:26,220 - ERROR - C PoW test fail.
Traceback (most recent call last):
  File "proofofwork.py", line 306, in init
  File "proofofwork.py", line 117, in _doCPoW
ValueError: Procedure probably called with too many arguments (12 bytes in excess)
2020-10-08 12:29:39,296 - WARNING - C:\Users\D.Bogomolov\AppData\Roaming\namecoin\namecoin.conf unreadable or missing, Namecoin support deactivated
2020-10-08 12:29:39,298 - WARNING - There was a problem testing for a Namecoin daemon. Hiding the Fetch Namecoin ID button
2020-10-08 12:29:39,298 - WARNING - No indicator plugin found
2020-10-08 12:29:39,298 - WARNING - No notification.message plugin found
2020-10-08 12:29:50,905 - WARNING - C:\Users\D.Bogomolov\AppData\Roaming\namecoin\namecoin.conf unreadable or missing, Namecoin support deactivated
2020-10-08 12:29:54,983 - WARNING - There was a problem testing for a Namecoin daemon. Hiding the Fetch Namecoin ID button
2020-10-08 12:40:53,221 - ERROR - UPnP send query failed
Traceback (most recent call last):
  File "upnp.py", line 319, in sendSearchRouter
error: [Errno 10049]

Hmm... Today it runs: ``` 2020-10-08 12:29:23,505 - WARNING - Using default logger configuration 2020-10-08 12:29:26,220 - ERROR - C PoW test fail. Traceback (most recent call last): File "proofofwork.py", line 306, in init File "proofofwork.py", line 117, in _doCPoW ValueError: Procedure probably called with too many arguments (12 bytes in excess) 2020-10-08 12:29:39,296 - WARNING - C:\Users\D.Bogomolov\AppData\Roaming\namecoin\namecoin.conf unreadable or missing, Namecoin support deactivated 2020-10-08 12:29:39,298 - WARNING - There was a problem testing for a Namecoin daemon. Hiding the Fetch Namecoin ID button 2020-10-08 12:29:39,298 - WARNING - No indicator plugin found 2020-10-08 12:29:39,298 - WARNING - No notification.message plugin found 2020-10-08 12:29:50,905 - WARNING - C:\Users\D.Bogomolov\AppData\Roaming\namecoin\namecoin.conf unreadable or missing, Namecoin support deactivated 2020-10-08 12:29:54,983 - WARNING - There was a problem testing for a Namecoin daemon. Hiding the Fetch Namecoin ID button 2020-10-08 12:40:53,221 - ERROR - UPnP send query failed Traceback (most recent call last): File "upnp.py", line 319, in sendSearchRouter error: [Errno 10049] ```

g1itch commented

2020-10-08 12:03:01 +02:00

(Migrated from github.com)

And Bitmessage_x64.exe from https://download2.bitmessage.org/tmp/20200928/ also runs on that machine.

PeterSurda commented

2020-10-08 14:49:03 +02:00

(Migrated from github.com)

The current 64bit binaries don't crash on Windows 10, but they can't send/receive messages because pyelliptic is broken. I still haven't tried running the binaries from this PR.

g1itch commented

2020-10-08 14:51:54 +02:00

(Migrated from github.com)

The current 64bit binaries don't crash on Windows 10, but they can't send/receive messages because pyelliptic is broken. I still haven't tried running the binaries from this PR.

I see. The date/status of sent message is "Waiting for their encryption key...".

> The current 64bit binaries don't crash on Windows 10, but they can't send/receive messages because pyelliptic is broken. I still haven't tried running the binaries from this PR. I see. The date/status of sent message is "Waiting for their encryption key...".

PeterSurda commented

2020-10-08 14:54:28 +02:00

(Migrated from github.com)

I see. The date/status of sent message is "Waiting for their encryption key...".

Yes, that's how the problem manifests. It can't decode the pubkey object.

> I see. The date/status of sent message is "Waiting for their encryption key...". Yes, that's how the problem manifests. It can't decode the pubkey object.

g1itch commented

2020-10-08 15:51:41 +02:00

(Migrated from github.com)

I see. The date/status of sent message is "Waiting for their encryption key...".

Yes, that's how the problem manifests. It can't decode the pubkey object.

i386 build behaves in the same way on my machine.

> > I see. The date/status of sent message is "Waiting for their encryption key...". > > Yes, that's how the problem manifests. It can't decode the pubkey object. i386 build behaves in the same way on my machine.

PeterSurda commented

2020-10-08 19:23:05 +02:00

(Migrated from github.com)

i386 build behaves in the same way on my machine.

It's possible the bug affects other combinations of OS and CPU. However, to get a full test, you have to empty the inventory table between trying different versions of the binary / code, or always use an empty config directory. Maybe you haven't done that when switching from 64bit to 32bit binary?

> i386 build behaves in the same way on my machine. It's possible the bug affects other combinations of OS and CPU. However, to get a full test, you have to empty the `inventory` table between trying different versions of the binary / code, or always use an empty config directory. Maybe you haven't done that when switching from 64bit to 32bit binary?

g1itch commented

2020-10-08 19:40:50 +02:00

(Migrated from github.com)

I removed messages.dat and created support request for all three exes: Bitmessage_x64 and _x86 from https://download2.bitmessage.org/tmp/20200928 and my build with these changes. The same result: nothing is sent.