I debated with myself whether to create the home_carousel relation or
rather if it would be better to have a single carousel relation for all
pages. However, i thought that it would be actually harder to maintain
a single relation because i would need an additional column to tell one
carrousel from another, and what would that column be? An enum? A
foreign key to another relation? home_carousel carries no such issues.
I was starting to duplicate logic all over the packages, such as the
way to encode media paths or “localization” (l10n) input fields.
Therefore, i refactorized them.
In the case of media path, i added a function that accepts rows of
media, because always need the same columns from the row, and it was
yet another repetition if i needed to pass them all the time. Plus,
these kind of functions can be called as `table.function`, that make
them look like columns from the table; if PostgreSQL implemented virtual
generated columns, i would have used that instead.
I am not sure whether that media_path function can be immutable. An
immutable function is “guaranteed to return the same results given the
same arguments forever”, which would be true if the inputs where the
hash and the original_filename columns, instead of the whole rows, but
i left it as static because i did not know whether PostgreSQL interprets
the “same row but with different values” as a different input. That is,
whether PostgreSQL’s concept of row is the actual tuple or the space
that has a rowid, irrespective of contents; in the latter case, the
function can not be immutable. Just to be in the safe side, i left it
stable.
The home page was starting to grow a bit too much inside the app
package, new that it has its own admin handler, and moved it all to a
separate package.
Debian’s pgtype version is 1.10.0, but pgtype did not add RecordArray
type until 1.11.0, thus could not compile the application with
Debian 12.
I just copied the code from 1.11.0 without any modification besides
adding the reference to the pgtype package to types and functions.
I am not happy with the localization interface for admins, but it is the
easier that i could think of (for me, i guess), with a separate for
each language.
I am not at all proud of the use of RecordArray, but i did not see the
need to create and register a type just to show the translation links.
I might change my mind when i need to add more and more translation
links, but only it the current interface remains, which i am not that
sure at the moment.
This is the image that is shown at the home page, and maybe other pages
in the future. We can not use a static file because this image can be
changed by the customer, not us; just like name and description.
I decided to keep the actual media content in the database, but to copy
this file out to the file system the first time it is accessed. This is
because we are going to replicate the database to a public instance that
must show exactly the same image, but the customer will update the image
from the private instance, behind a firewall. We could also synchronize
the folder where they upload the images, the same way we will replicate,
but i thought that i would make the whole thing a little more brittle:
this way if it can replicate the update of the media, it is impossible
to not have its contents; dumping it to a file is to improve subsequent
requests to the same media.
I use the hex representation of the media’s hash as the URL to the
resource, because PostgreSQL’s base64 is not URL save (i.e., it uses
RFC2045’s charset that includes the forward slash[0]), and i did not
feel necessary write a new function just to slightly reduce the URLs’
length.
Before checking if the file exists, i make sure that the given hash is
an hex string, like i do for UUID, otherwise any other check is going
to fail for sure. I moved out hex.Valid function from UUID to check for
valid hex values, but the actual hash check is inside app/media because
i doubt it will be used outside that module.
[0]: https://datatracker.ietf.org/doc/html/rfc2045#section-6.8
Seasons have a color to show on the calendar. I need them in HTML format
(e.g., #123abc) in order to set as value to `<input type="color">`, but
i did not want to save them as text in the database, as integers are
better representations of colors—in fact, that’s what the HTML syntax
also is: an integer.
I think the best would be to create an extension that adds an HTML color
type, with functions to convert from many representations (e.g., CSS’
rgb or even color names) to integer and back. However, that’s a lot of
work and i can satisfy Camper’s needs with just a couple of functions
and a domain.
To show the color on the index, at first tried to use a read-only
`<input type="color">`, but seems that this type of input can not be
read-only and must be disabled instead. However, i do not know whether
it makes sense to have a disabled input outside a form “just” to show
a color; i suspect it does not. Thus, at the end i use SVG with a
single circle, which is better that a 50%-rounded div with a background
color, even if the result is the same—SVG **is** intended for showing
pictures, which is this case.
It is inside the “user menu” only because this is where Numerus has the
same option, although it makes less sense in this case, because Numerus
is geared toward individual freelancers while Camper is for companies.
But, since it is easy to change afterward, this will do for now.
However, it should be only shown to admin users, because regular
employees have no UPDATE privilege on the company relation. Thus, the
need for a new template function to check if the user is admin.
Part of #17.
For now, there is only the label, type, and active fields. We will need
some field to hold the area on the map, but this requires #4, and
possibly #6, to be finished.
Part of #27.
At first i thought i had to return HTTP 410 gone in this case, but
HTTP Semantics RFC[0] says that “The 410 (Gone) status code indicates
that […] this condition is likely to be permanent. If the origin server
does not know […] whether or not the condition is permanent, the status
code 404 (Not Found) ought to be used instead.”
A non-active campsite type does not mean “deleted”, but rather
temporarily disabled, thus a 404 is the appropriate code.
[0] https://www.rfc-editor.org/rfc/rfc9110#status.410
GrapesJS was not working: too complex for users and not enough for
designers.
Therefore, we decided to use a simple WYSIWYG widget for the campsite
types’ description, while we will do the actual HTML template with an
external editor. Once that is done, we will convert that HTML to Go
templates and get the description’s content from the database.
Now the pages section has no sense: all the pages will be straight Go
templates. Only the pages for “special things”, like campsite types,
will use the database, and only for some fields, not the whole page.
I simply can not use Gutenberg without having it choking in its own
over-engineered architecture: using it inside a form, submits it when
clicking the button to change a paragraph’s text size; and using the
custom text size in pixels causes the paragraph component to fail.
The issue with paragraph’s custom text size is that block-editor’s
typography hook expects the font size to be a string, such as '12px' or
'1em', to call startsWith on it, but the paragraph sets an integer,
always assuming that the units are pixels. Integers do not have a
startsWith method.
Looking at the Gutenberg distributed with the current version of
WordPress, 6.3, seems that now paragraph has a selector for the units,
therefore never sets just the integer. That made me think that the
components used by the Isolated Block Editor are “mismatched”: maybe in
a previous version of block-editor it was always passed as an integer
too?
I downloaded the source code of the Isolated Block Editor and tried to
update @wordpress/block-library from version 8.14.0 to the current
version, 8.16.0, but fails with an error saying that 'core/paragraph' is
not registered, when, as far as i could check, it was. Seems that
something changed in @wordpress/blocks between version 12.14.0 and
12.16.0, so i tried to upgrade that module as well; it did not work
because @wordpress/data was not updated —do not remember the actual
error message—. Upgrading to @wordpress/data from 9.7.0 to 9.9.0 made
the registration of the 'isolated/editor' subregistry to be apparently
ignored, because the posterior select('isolated/editor') within a
withSelect hook returns undefined.
At this point, i gave up: it is obvious that the people that shit
JavaScript for Gutenberg do not care for semantic versioning, and there
are a lot of moving parts to fix just to be able to use a simple
paragraph block!
It seems, however, that there are not many open-source, block-based
_layout_ editors out there: mainly GrapesJS and Craft.JS. Craft.JS,
however, has no way to output HTML[0], requiring hacks such as using
React to generate the HTML and then pasted that shit onto the page;
totally useless for me.
I am not a fan of GrapesJS either: it seems that the “text block” is
a content-editable div, and semantic HTML can go fuck itself,
apparently. Typical webshit mentality. By strapping another huge
dependency like CKEditor, but only up to the already out-of-support
version 4, i can write headers, paragraphs and list. That’s
something, i guess.
[0]: https://github.com/prevwong/craft.js/issues/42
Part of #33.
For now, this is almost identical to the campsite types, but this
section is for purely informational pages that have no other relation
to the database than “belongs to the same company”.
Part of #33.
Had to export and move PublicPage struct to template because i can not
import app from campsites/types: app already imports campsite for the
http handler, and it, in turn, imports the types package for its own
http handler; an import loop.
Also had to replace PublicPage.MustRender with a Setup function because
the page passed down to html/template was the PublicPage struct, not
whatever struct embeds it. I was thinking more of Java inheritance here
rather than struct embedding.
I was starting to add the public page for campsite types, creating more
granular row-level security policies for select, insert, update, and
delete, because now the guest users needed to SELECT them and they have
no related company to filter the rows with. Suddenly, i realized that
the role was wrong in the user relation: a user can be an admin to one
company, and employee to another, and guess to yet another company;
the role should be in the company_user relation instead.
That means that to know the role to set to, the user alone is not enough
and have to know the company as well. Had to change all the
cookie-related function to accept also the company’s host name, as this
is the information that the Go application has.
Since the locales is a map, and maps in Go do not have order, sometime
the language switcher was shown in a different order.
I sort by language code, which is as arbitrary as sorting by name, but
makes sense to me.
I realized that locales should be company-dependent: we could have two
companies that show pages in a different subset of the application
locales. It is not the case now, because despite being a “multicompany
application”, it is intended for a single customer, but still makes
sense to include it in Company, even if the subset is the same set as
the application’s.
The language switcher needs the same information as languageLinks
needed, namely the list of locales and the current Path, to construct
the URI to all alternate versions. However, in this case i need access
to this data in the template context, to build the list of links.
At first i use request’s context to hold the list of available locales
from application, and it worked, possibly without ill-effects, but i
realized that i was doing it just to avoid a new parameter. Or, more
precise, an _explicit_ parameter; the context was used to skip the
inner functions between app and template.MustRenderPublic, but the
parameter was there all the same.
Finally, i thought that some handler might want to filter the list of
locales to show only the ones that it has a translation of. In that
case, i would need to extract the locales from the context, filter it,
and create a new request with the updated context. That made little
sense, and made me add the explicit locales parameter.
Since now the template has the same data as languageLinks, there is
little point of having the link in the HTTP response headers, and added
the <link> elements to <head>.
I thought that maybe i could avoid these <links> as they give the exact
same data as the language switch, but Google says nothing of using
regular anchors to gather information about localized versions of the
document[0], thus i opted to be conservative. One can reason that the
<head> has more weight for Google, as most sites with user-generated
content, which could contain these anchors, rarely allow users to edit
the <head>.
[0]: https://developers.google.com/search/docs/specialty/international/localized-versions
I was keeping this variable to redirect to the requested URL when the
user has no permission, is not logged in, and is shown the login form to
redirect them back to the original URL. Since each handler removes a
part of the path each time, i need to keep the original Path for that.
However, i just found out that request.RequestURI already is that
original URI requested by the client. No need for an extra variable.
I need to check that the user is an employee (or admin) in
administration handlers, but i do not want to do it for each handler,
because i am bound to forget it. Thus, i added the /admin sub-path for
these resources.
The public-facing web is the rest of the resources outside /admin, but
for now there is only home, to test whether it works as expected or not.
The public-facing web can not relay on the user’s language settings, as
the guest user has no way to set that. I would be happy to just use the
Accept-Language header for that, but apparently Google does not use that
header[0], and they give four alternatives: a country-specific domain,
a subdomain with a generic top-level domain (gTLD), subdirectories with
a gTLD, or URL parameters (e.g., site.com?loc=de).
Of the four, Google does not recommend URL parameters, and the customer
is already using subdirectories with the current site, therefor that’s
what i have chosen.
Google also tells me that it is a very good idea to have links between
localized version of the same resources, either with <link> elements,
Link HTTP response headers, or a sitemap file[1]; they are all
equivalent in the eyes of Google.
I have choosen the Link response headers way, because for that i can
simply “augment” ResponseHeader to automatically add these headers when
the response status is 2xx, otherwise i would need to pass down the
original URL path until it reaches the template.
Even though Camper is supposed to be a “generic”, multi-company
application, i think i will stick to the easiest route and write the
templates for just the “first” customer.
[0]: https://developers.google.com/search/docs/specialty/international/managing-multi-regional-sites
[1]: https://developers.google.com/search/docs/specialty/international/localized-versions
I had a weird but that sometimes the application would use Spanish as
the default language, even though the user had not passed Spanish as
value of the Accept-Language header and the default is hard-coded to
Catalan.
I learned that language.Matcher **always** returns one of its defined
locales, even if there is no way to match the passed list of languages
to Match. In that case, it returns a confidence of “No”, meaning that
the match failed.
Usually go the default Catalan locale because most of the time this was
set as the first language to Matcher, but since Go randomizes maps,
there were times that Spanish was first.
This form has an “HTML field”, which is just a <textarea> but “improved”
with the use of Automattic’s isolated block editor[0], a repackaged
Gutenberg’s editor playground as full-featured multi-instance editor
that does not require WordPress.
I do not want to use Node to build this huge, over-engineered piece of …
software. Therefore, i downloaded the released “browser” package, and
added the required React bundle, like i do with HTMx. This will hold
until i need a new custom block type; let’s hope i will not need it.
[0]: https://github.com/Automattic/isolated-block-editor
We made the decision that this application will also serve the public
pages to guests and customers, to avoid the overhead of having to
synchronize all data between this application and a bespoke WordPress
plugin.
That means that i no longer can have a /company/slug in the URL to know
which company the request is for, not only because it looks ugly but
because guest users do not have a “main company”—or any company
whatsoever.
Since the public-facing web is going to be served through a valid DNS
domain, and all companies are going to have a different domain, i
realized this is enough: i only had to add a relation of company and
their hosts. The same company can have many hosts for staging servers
or to separate the administration and public parts, for instance.
With change, the company is already known from the first handler, and
can pass it down to all the others, not only the handlers under
/company/slug/whatever. And i no longer need the companyURL function,
as there is no more explicit company in the URL.
Even though template technically does not need the template, as it only
contains the ID —the rest of the data is in a relation inaccessible to
guests for now—, but i left the parameter just in case later on i need
the decimal digits or currency symbol for whatever reason.
I really doubt that they are going to use more than a single company,
but the application is based on Numerus, that **does** have multiple
company, and followed the same architecture and philosophy: use the URL
to choose the company to manage, even if the user has a single company.
The reason i use the slug instead of the ID is because i do not want to
make the ID public in case the application is really used by employees
of many unrelated companies: they need not need to guess how many
companies there are based on the ID.
I validate this slug to be a valid UUID instead of relaying on the
query’s empty result because casting a string with a malformed value to
UUID results in an error other than data not found. Not with that
select, but it would fail with a function parameter, and i want to add
that UUID check to all functions that do use slugs.
I based uuid.Valid function on Parse() from Google’s uuid package[0]
instead of using regular expression, as it was my first idea, because
that function is an order of magnitude faster in benchmarks:
goos: linux
goarch: amd64
pkg: dev.tandem.ws/tandem/numerus/pkg
cpu: Intel(R) Core(TM) i5-6200U CPU @ 2.30GHz
BenchmarkValidUuid-4 36946050 29.37 ns/op
BenchmarkValidUuid_Re-4 3633169 306.70 ns/op
The regular expression used for the benchmark was:
var re = regexp.MustCompile("^[a-fA-F0-9]{8}-[a-fA-F0-9]{4}-4[a-fA-F0-9]{3}-[8|9|aA|bB][a-fA-F0-9]{3}-[a-fA-F0-9]{12}$")
And the input parameter for both functions was the following valid UUID,
because most of the time the passed UUID will be valid:
"f47ac10b-58cc-0372-8567-0e02b2c3d479"
I did not use the uuid package as is, even though it is in Debian’s
repository, because i only need to check whether the value is valid,
not convert it to a byte array. As far as i know, that package can not
do that.
Adding the Company struct into auth was not my intention, as it makes
little sense name-wise, but i need to have the Company when rendering
templates and the company package has templates to render, thus using
the company package for the Company struct would create a dependency
loop between template and company. I’ve chosen the auth package only
because User is also there; User and Company are very much related in
this application, but not enough to include the company inside the user,
or vice versa, as the User comes from the cookie while the company from
the URL.
Finally, had to move methodNotAllowed to the http package, as an
exported function, because it is used now from other packages, namely
campsite.
[0]: https://github.com/google/uuid
It does nothing, but i need it to discuss with Oriol.
Now there are more than a single package that requires shiftPath, so
i moved it to http and an exported function, ShiftPath.
Part of #25.
I do not see the profile image as an “integral part” of the user (i.e.,
no need for constraints), hence i do not want to store it in the
database, as would do for the identification image during check-in.
By default, i store the avatars in /var/lib/camper/avatars, but it is a
variable to allow packagers change this value using the linker.
This is also served as a test bed for uploading files to the server,
that now has a better interface and uses less resources that what i did
to Numerus.
Now the profile handler needs to keep a variable to know the path to the
avatars’ directory, thus i had to change it to a struct nested in app,
much like the fileHandler does. It still has to return the HandlerFunc,
however, as this function needs to close over the user and connection
variables.
Part of #7.
This is the first form that uses HTMx, and can not return a 400 error
code because otherwise HTMx does not render the content.
Since now there are pages that do not render the whole html, with header
and body, i need a different layout for these, and moved the common code
to handle forms and such a new template file that both layouts can use.
I also need the request in template.MustRender to check which layout i
should use.
Closes#7.
Conceptually, to logout we have to “delete the session”, thus the best
HTTP verb would be `DELETE`. However, there is no way to send a
`DELETE` request with a regular HTML form, and it seems that never will
be[0].
I could use a POST, optionally with a “method override” technique, but
i was planing to use HTMx anyway, so this was as good an opportunity to
include it as any.
In this application i am not concerned with people not having JavaScript
enabled, because it is for a customer that has a known environment, and
we do not have much time anyway. Therefore, i opted to forgo
progressive enhancement in cases like this: if `DELETE` is needed, use
`hx-delete`.
Unfortunately, i can not use a <form> with a hidden <input> for the
CSRF token, because `DELETE` requests do not have body and the value
should be added as query parameters, like a form with GET method, but
HTMx does the incorrect thing here: sends the values in the request’s
body. That’s why i have to use a custom header and the `hx-header`
directive to include the CSRF token.
Then, by default HTMx targets the triggered element for swap with the
response from the server, but after a logout i want to redirect the
user to the login form again. I could set the hx-target to button to
replace the whole body, or tell the client to redirect to the new
location. I actually do not know which one is “better”. Maybe the
hx-target is best because then everything is handled by the client, but
in the case of logout, since it is possible that i might want to load
scripts only for logged-in users in the future, i opted for the full
page reload.
However, HTMx does not want to reload a page that return HTTP 401,
hence i had to include the GET method to /login in order to return the
login form with a response of HTTP 200, which also helps when
reloading in the browser after a failed login attempt. I am not worried
with the HTTP 401 when attempting to load a page as guest, because
this request most probably comes from the browser, not HTMx, and it will
show the login form as intended—even though it is not compliant, since
it does not return the WWW-Authenticate header, but this is the best i
can do given that no cookie-based authentication method has been
accepted[1].
[0]: https://www.w3.org/Bugs/Public/show_bug.cgi?id=10671#c16
[1]: https://datatracker.ietf.org/doc/id/draft-broyer-http-cookie-auth-00.html
The locale is completely dependent on the user, as much as its email or
CSRF token, so it does not make much sense to have it in a separate
variable: for different users we might have to use different locales.
Also, this means one less variable to pass to handlers, that most of
them will need the user at some point or another (i.e., to render its
profile icon).
The thing is that i can not import `app.User` from the template package
because it would create an import cycle. I created the `auth` package
just because of that.
I thought that the login code would be better moved to the auth package
as well, but of course then i would reintroduce the import cycle between
auth and template.
To get the user from the database i have to set the cookie first, that
was already done in database.MustAcquire, but i thought they were too
far apart, even thought they are so related. So, the cookie, and thus
the role, is set when getting the user, that is actually the first thing
to do once the connection is acquired. However, that way the database
package has no knowledge of cookies, and the code that sets the cookie
and retrieves the user are next to each other.
I applied the same logic to the changes of locale.Match: it has not
business knowing that the accept language string comes from a request;
it only needs the actual string. Also, the TODO comment about getting
the user’s locale made no sense, now, because app already knows that
locale, so there is no need to pass the user to the locale package.
Getting the locale is done after retrieving the user from the database,
for the same reason the connection is Acquired as far up as possible:
almost every request will need this value, together with the user and
the database connection.
I am a bit affraid that i will end up with functions that always expect
these three values. Maybe i can put the locale inside user, as it is
the user’s locale, after all, no matter if it came from the database or
the user agent, but connection and user must be separate, i think.
We’ll see.
Almost all request will need the database connection, either because
they will perform queries or because they need to check if the user is
logged in, for instance, so it should be acquired as far above as
possible to avoid acquiring multiple connections.
This is specially true since i have to pass the cookie to the database
to switch role and set request.user.email and request.user.cookie
config variables. I do not want to do that many times per request.
It is a lot of code having to check the login variables inside the POST
handler, and i could not mark each input field individually as invalid
because the generic errors array i was using did no identify which field
had the error.
Thus, i use more or less the same technique as with Numerus: a struct
with the value and the error message. This time the input field does
not have the label and extra attributes because i believe this belongs
to the template: if i want do reuse the same form template, i should
create a common template rather than defining everything in Go.
The name is a bit different, however, because it has meaning both to the
front and back ends, as it needs to be exactly the same. Writing it
twice is error-prone, as with a rename i could easily forget to change
one or the other, and here i see value in having that in Go, because
it is also used in code.
xgettext does not recognize Get as marker for translatable strings, and
i can not add it as is because Go has many functions called Get, and it
would mark many strings as translatable that should not be, like HTTP
headers.
I believe that Gettext is unusual in Go, because the “correct” way to
spell that function in Go would be GetText, and would not interfere in
any other function.
GettextNoop is just a function that marks the string as translatable but
does not translate it at all. I want this for error messages and such,
that i have to pass the string to a validator function, but it is only
necessary to translate it when the validation fails. I doubt it makes
a difference in peformance, but still.
I now actually handle the /login URL and check whether the email and
password are valid, creating the session cookie if correct, but doing
nothing else with that cookie, for now.
The validation is done by hand for now, because i do not yet how i will
actually do it without so much duplication.
It does nothing more than to server a single page that does nothing
interesting.
This time i do not use a router. Instead, i am trying out a technique
i have seen in an article[0] that i have tried in other, smaller,
projects and seems to work surprisingly well: it just “cuts off” the
URI path by path, passing the request from handler to handler until
it finds its way to a handler that actually serves the request.
That helps to loosen the coupling between the application and lower
handlers, and makes dependencies explicit, because i need to pass the
locale, company, etc. down instead of storing them in contexts. Let’s
see if i do not regret it on a later date.
I also made a lot more packages that in Numerus. In Numerus i actually
only have the single pkg package, and it works, kind of, but i notice
how i name my methods to avoid clashing instead of using packages for
that. That is, instead of pkg.NewApp i now have app.New.
Initially i thought that Locale should be inside app, but then there was
a circular dependency between app and template. That is why i created a
separate package, but now i am wondering if template should be inside
app too, but then i would have app.MustRenderTemplate instead of
template.MustRender.
The CSS is the most bare-bones file i could write because i am focusing
in markup right now; Oriol will fill in the file once the application is
working.
[0]: https://blog.merovius.de/posts/2017-06-18-how-not-to-use-an-http-router/
jordi fita mas