Size of BOINC download

[Jord van der Elst]

Normal BOINC download size is 8.4, 8.6MB.

7.16.16 is 13MB. What causes this discrepancy?

Cheers,

-- Jord van der Elst.

[Vitalii Koshura]

There are some extra dlls that are not needed anymore but still exist in the installer. Will be fixed later.

Best regards,

Vitalii Koshura

ср, 28 окт. 2020 г. в 20:28, Jord van der Elst <els...@gmail.com>:

--
You received this message because you are subscribed to the Google Groups "boinc_alpha" group.
To unsubscribe from this group and stop receiving emails from it, send an email to boinc_alpha...@ssl.berkeley.edu.
To view this discussion on the web visit https://groups.google.com/a/ssl.berkeley.edu/d/msgid/boinc_alpha/CAEXjb0dd9Y_WvPw5Y1hCiHFf3oS3JUHm8ymcy-KhSJdjrM2Wiw%40mail.gmail.com.

[Jord van der Elst]

Not so sure about that.

With thanks to Richard for posting this elsewhere:

Contents of 7.16.11

02/09/2020  15:05         1,306,952 boinc.exe
02/09/2020  15:05           399,176 boinccmd.exe
02/09/2020  15:06         8,281,416 boincmgr.exe
02/09/2020  03:31         1,928,192 boincscr.exe
02/09/2020  15:05            16,712 boincsvcctrl.exe
02/09/2020  15:05            70,472 boinctray.exe
               6 File(s)     12,002,920 bytes

Contents of 7.16.16

26/10/2020  18:33         4,325,200 boinc.exe
26/10/2020  18:33           684,880 boinccmd.exe
26/10/2020  18:33        10,739,536 boincmgr.exe
26/10/2020  18:33         3,661,648 boincscr.exe
26/10/2020  18:33           129,360 boincsvcctrl.exe
26/10/2020  18:33           345,936 boinctray.exe
               6 File(s)     19,886,560 bytes

Almost everything got doubled to eight times the normal size.

The .dll files in my install directory are the normal size 345KB - 2,234KB. So purely the executables got increased in size exponentially.

But okay, if it gets fixed the next time around, swell. People might suspect this version though, that it has extra payload. My Kaspersky didn't yell at me, so I don't think so.

-- Jord van der Elst.

[Richard Haselgrove]

You mean the old static VS2010 runtime DLLs (that ought to have been replaced by common shared files which would have been automatically updated)?

Yes, they're there, but they don't account for the increase. The six .exe files alone are 7.5 MB bigger than their predecessors. I suppose we blame Microsoft?

------ Original Message ------
From: "Vitalii Koshura" <lestat.d...@gmail.com>
To: "Jord van der Elst" <els...@gmail.com>
Cc: "BOINC Alpha Mailing List" <boinc...@ssl.berkeley.edu>
Sent: Wednesday, 28 Oct, 20 At 19:48
Subject: Re: [boinc_alpha] Size of BOINC download

There are some extra dlls that are not needed anymore but still exist in the installer. Will be fixed later.

Best regards,

Vitalii Koshura

ср, 28 окт. 2020 г. в 20:28, Jord van der Elst <els...@gmail.com>:

Normal BOINC download size is 8.4, 8.6MB.

7.16.16 is 13MB. What causes this discrepancy?

Cheers,

-- Jord van der Elst.

--
You received this message because you are subscribed to the Google Groups "boinc_alpha" group.

To unsubscribe from this group and stop receiving emails from it, send an email to unsub...@ssl.berkeley.edu" target="_blank">boinc_alpha+unsub...@ssl.berkeley.edu.

To view this discussion on the web visit https://groups.google.com/a/ssl.berkeley.edu/d/msgid/boinc_alpha/CAEXjb0dd9Y_WvPw5Y1hCiHFf3oS3JUHm8ymcy-KhSJdjrM2Wiw%40mail.gmail.com.

--

You received this message because you are subscribed to the Google Groups "boinc_alpha" group.

To unsubscribe from this group and stop receiving emails from it, send an email to unsub...@ssl.berkeley.edu">boinc_alpha+unsub...@ssl.berkeley.edu.
To view this discussion on the web visit https://groups.google.com/a/ssl.berkeley.edu/d/msgid/boinc_alpha/CAECimMWgVHPyJn94a721CaQYyscF6qd%3DUa6%3DQN0tY9UzTdtijA%40mail.gmail.com.

[Vitalii Koshura]

Because I made static linkage instead of dynamic (to get rid of dlls and make it unified: before some components were linked statically, some dynamically and it was not obvious).

Best regards,

Vitalii Koshura

ср, 28 окт. 2020 г. в 21:16, Jord van der Elst <els...@gmail.com>:

[Vitalii Koshura]

Just curious: does this really important to you? We're not in the early 90th to count every Kb ;)

If you want, I can't explain for every binary exactly what changed and what it includes now in comparison with the previous release :)

Best regards,

Vitalii Koshura

ср, 28 окт. 2020 г. в 21:18, Richard Haselgrove <r.hase...@btopenworld.com>:

[Richard Haselgrove]

Not me - people like me pony up for unrestricted access.

But people in countries where metered connections are the norm, or relying on mobile connections when off-grid. If we're asking people to volunteer, it's polite and considerate to think about things like that.

To view this discussion on the web visit https://groups.google.com/a/ssl.berkeley.edu/d/msgid/boinc_alpha/CAECimMUs%2Boq7ws0kPTtsEvKKPFx4xi4HWruBM76Y5rXQM7bpFA%40mail.gmail.com.

[Jord van der Elst]

Plus, since we have bugs we don't know if those are caused by the program size or something else.

And with years of the BOINC installer being 8 to 9 MB (https://boinc.berkeley.edu/dl/?C=M;O=D) and all of a sudden it's 13MB and the main developer waves it off with "is that something you find important?", then alarm bells start ringing.

Consistency is important.

-- Jord van der Elst.

[Vitalii Koshura]

Damn, it's just a test build with updated build tools, updated dependencies.
Yes, it differs from the previous one because it's built with VS2019 not VS2010 as before. Yes, binary size differs because we updated dependencies and made it more consistent (all binaries are built the same way. And I told about this before: https://github.com/BOINC/boinc/pull/3504. Yes, I forgot to remove the installer dlls that are not used anymore. It's my fuck-up, I agree. Final release installer will be smaller than now but a little bit bigger than before. And I not just `wave it off`. I asked why this really matters for you? Richard (?) mentioned about users with limited connections. Ok, I understand this, but we didn't increase the size of the installer 10 times or more.

I hope I answered all your questions.

Best regards,

Vitalii Koshura

ср, 28 окт. 2020 г. в 21:36, Jord van der Elst <els...@gmail.com>:

[Jacob Klein]

Hi. He saw something unexpected, and called it out, for which I am thankful. And it helps to know the reasoning and expectation, which is also helpful. So, it sounds like our tool chain is changing a bit. That is also helpful to know.

Please also let us know when the size is what you expected, and we can test to make sure it still installs and runs correctly. Thanks.

It's basically all about expectations. Oh, and size matters ;)

---

From: Vitalii Koshura <lestat.d...@gmail.com>
Sent: Wednesday, October 28, 2020 4:50:38 PM

To: Jord van der Elst <els...@gmail.com>

Cc: Richard Haselgrove <r.hase...@btopenworld.com>; BOINC Alpha Mailing List <boinc...@ssl.berkeley.edu>

To unsubscribe from this group and stop receiving emails from it, send an email to boinc_alpha...@ssl.berkeley.edu.
To view this discussion on the web visit https://groups.google.com/a/ssl.berkeley.edu/d/msgid/boinc_alpha/CAECimMXAN1iWv76J-Arp8OJ1%2BdOUCLJhjzPH0xLfePVrVKvjng%40mail.gmail.com.

[Richard Haselgrove]

Yes, it was me that mentioned the download size in relation to metered connections. And yes, I agree the absolute size increase - in the context of an application which is designed to download science continuously for years to come - is trivial.

But yes, it was also me that started this whole issue running, simply by noticing a new file appearing in the public /dl directory without explanation.

I still think that was wrong. I'm more than willing - as you've seen over the years - to participate in whatever tests are necessary. But tests have their purposes and their place. The designated place for early-stage alpha testing is the master branch of GitHub, and your help in extending and maintaining the automated testing tools there is much appreciated. Thank you.

But the release branch, and the public download directory, should be reserved for:

a) Final beta testing of release candidate builds

b) Clearly identified private drops of apps required to test special cases.

If anyone can think of another valid use, please shout. But if no further cases are agreed, I think we should stick to that formulation.

To view this discussion on the web visit https://groups.google.com/a/ssl.berkeley.edu/d/msgid/boinc_alpha/CAECimMXAN1iWv76J-Arp8OJ1%2BdOUCLJhjzPH0xLfePVrVKvjng%40mail.gmail.com.