MIME-Version: 1.0
References: <99A6E7C8-1634-4162-96FD-D5AC404ADBCF@gmx.net>
 <29c71778-5871-4e4d-a80e-d67351c8eb4b@postgresql.org>
 <4C210077-20E5-415C-8A69-B82331A42B9A@gmx.net>
In-Reply-To: <4C210077-20E5-415C-8A69-B82331A42B9A@gmx.net>
From: Magnus Hagander <magnus@hagander.net>
Date: Mon, 8 Sep 2025 10:50:46 +0200
Message-ID: 
 <CABUevEz+R64pT9PZJB6Lq=S4AQA4hQFyXZuneDKva2MGEfXZUA@mail.gmail.com>
Subject: Re: Images in news mails
To: Tobias Bussmann <t.bussmann@gmx.net>
Cc: "Jonathan S. Katz" <jkatz@postgresql.org>, pgsql-www@lists.postgresql.org
Content-Type: multipart/alternative; boundary="000000000000513d6e063e464906"
Archived-At: 
 <https://www.postgresql.org/message-id/CABUevEz%2BR64pT9PZJB6Lq%3DS4AQA4hQFyXZuneDKva2MGEfXZUA%40mail.gmail.com>
Precedence: bulk

--000000000000513d6e063e464906
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Sun, Sep 7, 2025 at 6:16=E2=80=AFPM Tobias Bussmann <t.bussmann@gmx.net>=
 wrote:

> Am 05.09.2025 um 20:23 schrieb Jonathan S. Katz <jkatz@postgresql.org>:
> > This has plagued me for years, and I'm so happy you tracked this down!
>
> I'm happy to hear, I'm not the only one seeing this. Glad to help!
>

What's more amazing is that AFAIK this is the first time *anyone* has
reported this to be a problem. And it's been this way since 2020...

It is also slightly ironic that it's apple mail that's seeing the problem,
given it's quite overrepresented in *generating* broken attachments in
similar ways when you look in the archive :)


> The patch looks good; have you tested in a few other mail clients? I
> haven't had a chance to test yet.
>
> I haven't tried other mail clients myself, but as the style definition is
> just redundant to the img attributes, I wouldn't expect any regression by
> this. But HTML emails are full of miracles ;)
>

Indeed it does. For example, you mentioned the possibility of embedding the
image instead of attaching - that *does* break multiple clients, I think at
least Outlook and maybe also gmail.


> Litmus.com has become quite expensive for render previews, so I looked fo=
r
> alternatives. I found Testi.at, which offers some free tests at least.
> Unfortunately, I was unable to provide the full MIME structure there and
> thus could not simulate the CID-referenced image that causes the problem =
in
> Apple Mail. When I experimented with an externally hosted image, the resu=
lt
> looked good in the provided previews. However, I visually identified
> another issue: Outlook does not seem to support the 'float' style attribu=
te
> [1 footnote 1]. In the v2 patch, I implemented a workaround to duplicate
> the 'float' style using an 'align' img attribute. Previews of the resulti=
ng
> rendering can be found at [2].
>
>
PGEU does have a subscription to a service that could be used for testing.
We'd need to adapt the test instance of the website for it, but that's
fairly easy to do.

I'll also see if I can dig out my old notes for how we ended up in this
stricture in the first place, because it did go back and forth a bunch
before actually merged.

//Magnus

--000000000000513d6e063e464906
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr"><br></div><br><div class=3D"gmail_quote g=
mail_quote_container"><div dir=3D"ltr" class=3D"gmail_attr">On Sun, Sep 7, =
2025 at 6:16=E2=80=AFPM Tobias Bussmann &lt;<a href=3D"mailto:t.bussmann@gm=
x.net">t.bussmann@gmx.net</a>&gt; wrote:<br></div><blockquote class=3D"gmai=
l_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,20=
4,204);padding-left:1ex">Am 05.09.2025 um 20:23 schrieb Jonathan S. Katz &l=
t;<a href=3D"mailto:jkatz@postgresql.org" target=3D"_blank">jkatz@postgresq=
l.org</a>&gt;:<br>
&gt; This has plagued me for years, and I&#39;m so happy you tracked this d=
own!<br>
<br>
I&#39;m happy to hear, I&#39;m not the only one seeing this. Glad to help!<=
br></blockquote><div><br></div><div>What&#39;s more amazing is that AFAIK t=
his is the first time *anyone* has reported this to be a problem. And it=
9;s been this way since 2020...</div><div><br></div><div>It is also slightl=
y ironic that it&#39;s apple mail that&#39;s seeing the problem, given it&#=
39;s quite overrepresented in *generating* broken attachments in similar wa=
ys when you look in the archive :)</div><div><br></div><div><br></div><bloc=
kquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:=
1px solid rgb(204,204,204);padding-left:1ex">
&gt; The patch looks good; have you tested in a few other mail clients? I h=
aven&#39;t had a chance to test yet.<br>
<br>
I haven&#39;t tried other mail clients myself, but as the style definition =
is just redundant to the img attributes, I wouldn&#39;t expect any regressi=
on by this. But HTML emails are full of miracles ;)<br></blockquote><div><b=
r></div><div>Indeed it does. For example, you mentioned the=C2=A0possibilit=
y of embedding the image instead of attaching - that *does* break multiple =
clients, I think at least Outlook and maybe also gmail.</div><div><br></div=
><div><br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0=
px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
Litmus.com has become quite expensive for render previews, so I looked for =
alternatives. I found Testi.at, which offers some free tests at least. Unfo=
rtunately, I was unable to provide the full MIME structure there and thus c=
ould not simulate the CID-referenced image that causes the problem in Apple=
 Mail. When I experimented with an externally hosted image, the result look=
ed good in the provided previews. However, I visually identified another is=
sue: Outlook does not seem to support the &#39;float&#39; style attribute [=
1 footnote 1]. In the v2 patch, I implemented a workaround to duplicate the=
 &#39;float&#39; style using an &#39;align&#39; img attribute. Previews of =
the resulting rendering can be found at [2].<br><br></blockquote><div><br><=
/div><div>PGEU does have a subscription to a service that could be used for=
 testing. We&#39;d need to adapt the test instance of the website for it, b=
ut that&#39;s fairly easy to do.</div><div><br></div><div>I&#39;ll also see=
 if I can dig out my old notes for how we ended up in this stricture in the=
 first place, because it did go back and forth a bunch before actually merg=
ed.</div><div><br></div><div>//Magnus</div><div>=C2=A0</div></div><div dir=
=3D"ltr" class=3D"gmail_signature"><div dir=3D"ltr"></div></div></div>

--000000000000513d6e063e464906--